(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Biodiversity Heritage Library | Children's Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "Role of cis- and trans- elements in mitochondrial transcription"

THE ROLE OF CIS- AND TRANS- ELEMENTS 
IN MITOCHONDRIAL TRANSCRIPTION 



By 

STEVEN C. GHIVIZZANI 



A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL 

OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT 

OF THE REQUIREMENTS FOR THE DEGREE OF 

DOCTOR OF PHILOSOPHY 

UNIVERSITY OF FLORIDA 

1991 



To my wife, Nancy. 



ACKNOWLEDGEMENTS 

I wish to thank my mentor, Dr. William Hauswirth, for 
the opportunity to work in his laboratory and his guidance 
and friendship during my graduate training. 

My appreciation is extended to all the faculty in the 
Department of Immunology and Medical Microbiology, 
particularly my committee members, Drs. Donna Duckworth, 
Henry Baker, and Edward Wakeland, as well as Dr. Al Lewin 
for their advice and assistance. A special thanks is 
extended to Dr. Parker Small for his advice, guidance, and 
friendship. 

I also wish to thank both past and present members of 
Dr. Hauswirth' s laboratory for their assistance and 
camaraderie, especially Drs. Cort Madsen, Douglas McCarty, 
Gery Hehman, Rein Hoedemakers, Andrew Hertz, Finn Pond, and 
Mary Ashley. I extend my most sincere appreciation to 
Chrissie Street for her friendship, support, and help 
throughout my entire graduate education. 

I wish to thank my mother, Carolyn Scheider; sister, 
Darling Fournier; grandparents, Jesse and Ceceila Charles, 
and especially my brother, Dr. Scot Ghivizzani for their 
constant encouragement and support. I would also like to 
thank Drs. David Nickerson, George Yarko, and Steven 
Poorbaugh for their years of friendship and encouragement. 

iii 



Lastly and most of all, I would like to thank my wife, 
Nancy, for all her help throughout my graduate education. 
Without her continuous support and companionship this work 
would not have been possible. 



iv 



TABLE OF CONTENTS 

page 

ACKNOWLEDGEMENTS iii 

ABSTRACT V 

CHAPTERS 

1 INTRODUCTION AND BACKGROUND 1 

2 MATERIALS AND METHODS 18 

3 SITES OF ARTIODACTYL L-STRAND RNA INITIATION... 33 

Introduction 33 

Results 35 

Discussion 57 

4 ISOLATION AND CHARACTERIZATION OF BOVINE AND 
PORCINE MITOCHONDRIAL RNA POLYMERASE ACTIVITIES 68 

Introduction 68 

Results 71 

Discussion 99 

5 ISOLATION OF A BOVINE mitTF ACTIVITY AND 
CHARACTERIZATION OF ITS DNA-BINDING 

SPECIFICITY 107 

Introduction 107 

Results 110 

Discussion 137 

6 INTERACTION OF BOVINE MITOCHONDRIAL RNA 
POLYMERASE AND TRANSCRIPTION FACTOR AT 

SITES OF TRANSCRIPTION INITIATION 148 

Introduction 148 

Results 151 

Discussion 170 

7 SUMMARY AND PERSPECTIVES 181 

LIST OF REFERENCES 191 

BIOGRAPHICAL SKETCH 197 

V 



Abstract of Dissertation Presented to the Graduate School 

of the University of Florida in Partial Fulfillment of the 

Requirements for the Degree of Doctor of Philosophy 

THE ROLE OF CIS- AND TRANS- ELEMENTS 
IN MITOCHONDRIAL TRANSCRIPTION 

By 

Steven C. Ghivizzani 

December, 1991 

Chairperson: William W. Hauswirth 

Major Department: Immunology and Medical Microbiology 

To better define the DNA sequence elements necessary 

for regulation of mitochondrial RNA synthesis, in vivo and 

in vitro L-strand transcription were studied in several 

species within the Artiodactyl order. To test the 

functional relevance of conserved L-strand promoter domains, 

in vitro transcription systems were developed from bovine 

and porcine mitochondria. Although significant promoter 

sequence differences exist between species, the two RNA 

polymerase preparations were functionally interchangeable. 

Extending these studies, separable bovine mitochondrial 

transcription factor (mitTF) and RNA polymerase activities 

were identified chromatographically. Using DNAse 1 

protection assays, sites of specific bovine mitTF: DNA 

interaction were located within heavy and light strand 

promoters (HSP and LSP respectively) of the cow, as well as 

vi 



within LSP and HSP of the human and mouse and within the LSP 
of the pig. A sequence alignment of promoters bound by 
bovine mitTF suggests only general DNA sequence motifs are 
required for recognition. A synthetic Artiodactyl LSP 
region was made, incorporating conserved promoter domains 
separated by restriction enzyme sites, thus forming movable 
DNA cassettes. When tested by in vitro transcription and 
DNAse 1 protection, the bovine transcriptional activity 
unexpectedly was able to transcribe templates in which 
specific mitTF binding was nearly absent. When promoter 
domains were deleted entirely, transcription initiated at a 
nearby vector T3 promoter. These experiments suggest a 
dynamic relationship between mitTF and RNA polymerase. At 
limiting mitTF, transcription is confined to start sites 
near high affinity factor binding domains within LSP. At 
increased levels, factor binding occurs at lower affinity 
binding sites, allowing the RNA polymerase to select 
multiple sites of initiation. These results lead to a model 
for regulating the balance between RNA initiation leading to 
priming of DNA replication and RNA initiation leading to 
transcription of the mitochondrial genome. 



Vll 


















CHAPTER 1 
INTRODUCTION AND BACKGROUND 



Organization of the Mitochondrial Genome 

Mitochondria are cytoplasmic organelles primarily 
responsible for ATP production and fatty acid metabolism in 
aerobic eukaryotic cells. They can vary in number from a few 
to several hundred to nearly ten thousand depending on the 
type of tissue. The mitochondrion is defined by a double 
membrane which encloses a fluid matrix containing 
mitochondrial DNA (mitDNA) and the mitochondrially specific 
transcription and translation machinery. The inner membrane 
is highly enfolded and contains many of the proteins 
involved in oxidative phosphorylation. The outer membrane 
is smooth and contains various enzymes including those 
involved in fatty acid activation. An interesting feature 
of these organelles is their extranuclear genetic system. 
While all the processes involved in ATP synthesis occur 
within the mitochondrion, its genome only encodes a fraction 
of the proteins involved in oxidative phosphorylation as 
well as the mitochondrial tRNAs and rRNAs necessary for 
their translation (Anderson et al. 1981; Anderson et al. 
1982; Bibb et al. 1981). All other proteins, including 
those involved in mitDNA replication and transcription, are 

1 



encoded in the nucleus and imported from the cytoplasm 
(reviewed in Clayton 1984) . These features make the 
mitochondrion a unique system in which to study nuclear- 
organellar communication. 

The complete mitDNA sequences of the human, mouse, 
bovine, and Xenopus systems have been determined (Anderson, 
et al. 1981; Anderson et al. 1982; Bibb et al. 1981; Roe et 
al. 1985) . These genomes exhibit the identical gene 
organization, as well as significant sequence similarities 
within respective coding regions. The characteristic 
vertebrate mitochondrial genome is a double-stranded closed 
circular DNA molecule about 16.5 kb in length (Anderson, et 
al. 1981; Anderson et al. 1982; Bibb et al. 1981; Roe et al. 
1985) (diagrammed in Figure 1-1) . Differences in the G + T 
content of the two strands allow for their separation in 
CsCl gradients . Both "light" (L-) and "heavy" (H-) strands 
contain coding regions; however, the majority are located on 
the heavy strand. These coding regions are arranged on the 
genome in a fashion such that one or more tRNA genes are 
usually situated between each protein and rRNA coding 
region. In contrast to the 76 kb yeast mitochondrial genome 
in which 50-55% of the molecule is non-coding DNA, the 
vertebrate genome is very compact. Coding regions contain 
no introns and are contiguous, in most cases with no or very 
few bases of intergenic sequence. There is only one 
significant noncoding region, a stretch of about 1000 bases 



3 

which contains a triple stranded DNA segment known as the 
displacement or D-loop (diagrammed in Figure 1-1, Panel B) . 
While a comparison of DNA sequences from the vertebrates 
studied shows extensive sequence similarity within their 
coding regions, vertebrate D-loop DNA sequences are very 
dissimilar, with the exception of some short conserved 
sequence blocks (CSBs) and a region in the center of the D- 
loop (Anderson et al. 1982) • 

Mitochondrial Transcription 

The current model for vertebrate mitochondrial 
transcription suggests that RNA synthesis from both DNA 
strands initiates from two sites within the 5' end of the D- 
loop region and proceeds in opposite directions transcribing 
the entire genome (Aloni and Attardi 1971a; Murphy et al. 
1975; Montoya et al. 1982; reviewed in Clayton 1984). The 
resulting polycistronic RNA molecules are then processed 
into the individual mitochondrial RNAs. The spacing of the 
tRNA genes between mitochondrial protein and rRNA coding 
regions suggests their possible role as recognition 
structures for RNA processing enzymes. While there is only 
limited direct experimental evidence for this scheme several 
lines of indirect evidence form the basis for this proposed 
type of transcriptional organization. First, the 
mitochondrial genome is organized such that the absence of 
noncoding DNA between genes allows no apparent room for 






Figure 1-1. Map of the Mammalian Mitochondrial Genome and 
D-Loop Region . Panel A. Diagram of the 16.5 kb mammalian 
mitochondrial genome (Watson et al. 1987). Differences in 
the G + T content of the two DNA strands allow for their 
separation in cesium chloride gradients into heavy (H-) and 
light (L-) chains. Characteristic of mammalian 
mitochondrial genomes is the triple stranded DNA region 
indicated as the D-loop. The outer circle represents genes 
transcribed from the H-strand DNA, and the inner circle 
those genes transcribed from the L-strand. The origins of 
H- and L-strand replication are designated OH and OL 
respectively, with arrows indicating the direction of DNA 
synthesis. Heavily shaded regions indicate the 
mitochondrial 16S and 12S rRNA genes, while lighter shaded 
areas represent protein coding genes. Cytochrome oxidase 
subunits are shown as CO I, II, and III. Components of the 
respiratory chain NADH dehydrogenase are shown as ND 1-5. 
Cytochrome b is designated cyt.b. Unidentified reading 
frame 6 is designated as URF6. Black dots designate the 
tRNA genes. Panel B. Expanded diagram of the mammalian 
mitochondrial D-loop region. A detailed representation of 
the D-loop region in Panel A is shown in the opposite 
orientation. LSP and HSP indicate light and heavy strand 
promoter regions respectively, with arrows indicating 
directions of RNA synthesis. Dark arrows represent the 
primary direction of RNA synthesis, while light arrows 
indicate bidirectional transcription (see text) . The 
lightly shaded ellipses at these regions represent binding 
by the mitochondrial transcription factor (mitTF) . Blocks 
of sequence conserved among non-Artiodactyl species and the 
pig are indicated by CSBs -1, -2, and -3. As in Panel A, 
the 5' end of the D-loop DNA strand, the origin of H-strand 
replication, is designated OH. The arrow at the 3' end of 
the D-loop DNA indicates the direction of DNA synthesis. 
Termination associated sequences, regions believed to be 
important to the formation of the D-loop 5' end, are shown 
as TAS. Nearby tRNA genes for proline (PRO) , phenylalanine 
(PHE) , and leucine (LEU) are shown as indicated as are 
portions of the 12 S and 16S rRNA genes. The darkened 
ellipse shown between the 16S rRNA and the tRNALEU genes 
represents binding of the mTERF protein in this region (see 
text) . 



B. 






H-strand 




Leu(CUN) 
er (AGY) 
His 



ATPase 8 



_£ 



OH 



El 



"^ — I LSP lEPr— ^ 12S rRNA <flfc 



IRNA PRO TAS 



CSBs 1 2 -3 



IRNA PHE 1GS tRNA IRNA LEU 






individual regulatory sequences upstream of each coding 
region. Thus, separate initiation events for each gene seem 
unlikely unless they are contained intragenically (Clayton 
1984) . Second, RNA mapping studies have shown the existence 
of "cappable" light and heavy strand RNA species that 
originate from two nearby but separate regions between the 
5 ' end of the D-loop DNA strand and the beginning of the 
phenylalanine tRNA (see below) (Yoza et al. 1984) . 
Furthermore, the isolation of large duplex RNAs 
approximately 70% of genome length, as well as detection of 
genomic length DNA/RNA hybrids for light strand transcripts 
(Montoya et al. 1982), also support this type of 
transcriptional regulation. These observations, though, do 
not exclude the possibility of transcription initiation 
elsewhere in the genome. One feature inconsistent with this 
proposed model is the existence of at least one cappable RNA 
5 1 end which maps to a position immediately upstream of the 
human 12S rRNA gene (Montoya et al. 1983; Yoza and 
Bogenhagen 1984) , suggesting that transcription initiation 
is not wholly localized within the D-loop region. 
Additionally, RNA synthesis in yeast mitochondria has been 
shown to initiate from at least 2 different sites at 
positions upstream from the large and small rRNAs, 
mitochondrially encoded proteins, and tRNAs (Christianson 
and Rabinowitz 1983) . 



7 
Light Strand Transcription and Mitochondrial DNA Replication 

Studies of transcriptional regulation in vertebrate 
mitochondria have focused primarily on the D-loop region 
(diagrammed in Figure 1-1, Panel B) where detailed studies 
in several vertebrate species have detected prominent L- 
strand and H-strand RNAs whose 5 ' ends map to evolutionally 
conserved positions (Murphy 1975; Montoya 1982; Yoza and 
Bogenhagen 1984) . L-strand RNAs map to a region upstream of 
the origin of heavy strand replication (OH) and CSBs-1, -2, 
and -3 and downstream from the phenylalanine tRNA gene 
(Murphy 1975; Montoya 1982; Yoza 1984). S-l nuclease 
protection assays and in vitro capping studies of L-strand 
RNAs in the human (Yoza et al. 1984) and the frog ( Xenopus ) 
(Bogenhagen et al 1986) , confirm that RNA synthesis 
initiates from this site. In addition to transcription 
of the genome, the position of the light strand promoter 
upstream from the heavy strand replication origin suggests 
that RNA synthesis from this site may also be responsible 
for regulating the replication of mitochondrial DNA (Gillum 
1979; Chang and Clayton 1985; Chang et al. 1985). Although 
the mechanism of replication of the mitochondrial genome is 
not yet fully understood, it is thought to be primed by a 
transcription event initiating from the light strand 
promoter (LSP) which proceeds downstream to a series of 
sequences (the CSBs) that provide signals for the transition 
from RNA to heavy strand DNA synthesis (Chang et al. 1985). 



8 
Replication then either proceeds unidirectionally around the 
genome or is aborted about 500 to 700 nucleotides downstream 
near a set of termination associated sequences (TAS) , thus 
generating the triple stranded D-loop. The 5' end of this 
displacement DNA strand is accordingly termed the origin of 
heavy strand replication (OH) . If heavy strand replication 
is not halted at the TAS region, it proceeds around the DNA 
molecule, displacing the parental heavy strand. After about 
two thirds of the heavy strand has been displaced, the 
single stranded origin of light strand replication (OL) 
becomes exposed, allowing its stem-loop structure to form 
(Wong and Clayton 1985a) . This secondary structure then 
serves as a signal to a mitochondrial DNA primase/polymerase 
activity (distinct from the mitochondrial RNA polymerase) 
which then synthesizes a short RNA primer for unidirectional 
light strand replication (Wong and Clayton 1985b) . Steps 
involved in the resolution of daughter molecules have not 
yet been studied. 

Heavy Strand Transcription 

As with studies involving L-strand RNAs, H-strand RNA 
start sites have been mapped to a conserved position at the 
5' end of the D-loop region (Figure 1-1, Panel B) , 
immediately upstream of the phenylalanine transfer RNA and 
12S and 16S ribosomal RNA genes (Montoya et al 1983; Yoza et 
al. 1984; Chang et al. 1986). RNA synthesis initiating from 



9 

this site is believed to be responsible for transcription of 
the entire heavy strand DNA which contains both rRNA genes 
and the majority of the protein coding regions (Murphy et 
al. 1975). Recent investigations have found a protein 
(Mitochondrial Termination Factor: mTERF) in human 
mitochondrial extracts which promotes in vitro termination 
of heavy strand transcription at the junction between the 
16S rRNA gene and the leucine tRNA gene (Kruse et al. 1989; 
Hess et al. 1991) . The process of termination correlates 
with specific binding of the mTERF protein to DNA seguences 
at the junction of the 16S rRNA and leucine tRNA genes 
(Figure 1-1, Panel B) . A similar type of process has also 
been proposed for the termination of D-loop DNA synthesis, 
where protein interaction at or near the TAS seguences leads 
to formation of the 3' end of the D-loop DNA (G. Hehman, 
personal communication) . 

In Vitro Transcription 

Comparative analysis of the DNA seguences flanking the 
H- and L-strand transcription promoter sites in several 
species shows that these regions are very heterogenous, with 
respect to each other and among species. This seguence 
heterogeneity has prevented identification of seguence 
elements relevant to mitochondrial transcriptional 
regulation by their evolutionary conservation. Due to the 
lack of an efficient method for transforming foreign DNAs 



10 

into animal mitochondria, in vitro transcription systems 
have been developed, first for the human (Walberg and 
Clayton 1983) and later for mouse (Chang and Clayton 1986a) 
and Xenopus (Bogenhagen and Yoza 1986) , in an attempt to 
determine the specific DNA sequences responsible for 
controlling mitochondrial RNA synthesis. These 
investigations involved the isolation of an RNA polymerase 
activity from purified mitochondria and analysis of its 
ability to transcribe in vitro a series of defined DNA 
templates. Deletion mutagenesis studies in these species, 
using cloned DNAs which contained various portions of D-loop 
sequences, have demonstrated the existence of autonomous 
light (LSP) and heavy strand promoter (HSP) elements (Chang 
and Clayton 1984; Chang and Clayton 1986a; Chang and Clayton 
1986b; Bogenhagen and Romanelli 1988) , diagrammed in Figure 
1-1, panel B. In the mouse and human systems, the nature of 
these two transcriptional regulatory regions appears to be 
somewhat different. Although both LSP and HSP regions are 
recognized by the same RNA polymerase preparation 
(Bogenhagen and Yoza 1984; Chang and Clayton 1986a; Chang 
and Clayton 1986b) , in vitro initiation is significantly 
stronger at the LSP. Experiments using a series of 
templates with processively longer 3 ' deletions (relative to 
the direction of RNA synthesis) have shown that both 
promoters require only a few bases of mitochondrial sequence 
downstream from each initiation site for wild type levels of 



11 

expression (Bogenhagen and Yoza 1984; Chang and Clayton 
1986a; Chang and Clayton 1986b) . Significant reductions in 
transcriptional efficiency were noted when sequences 
upstream (5 1 ) from the start sites were removed. At the 
HSP, deletions of mitochondrial sequences to within about 15 
nucleotides of the H-strand start site were associated with 
only moderate reduction in transcription (Bogenhagen and 
Yoza 1984; Chang and Clayton 1984; Chang and Clayton 1986c). 
However, while deletions to similar regions upstream of the 
L-strand start site could still support accurate initiation, 
a severe reduction in transcription was detected 
(Bogenhagen and Yoza 1984; Chang and Clayton 1984; Chang and 
Clayton 1986a; Chang and Clayton 1986b) . A comparison of 
the DNA sequences required for accurate and efficient 
transcription from the two promoter elements showed that 
both regions are largely heterologous (Chang and Clayton 
1984) . Further experiments in the human system using a 
series of point mutations directed at or near the initiation 
sites demonstrated that a range of "core" promoter sequences 
could be recognized by the mitochondrial RNA polymerase in 
vitro (Hixson and Clayton 1985) . Analogous studies in 
Xenopus showed no apparent difference in light and heavy 
strand promoters, both consisting of relatively short DNA 
regions (18 bp) with dyad symmetry about their respective 
RNA start sites (Bogenhagen and Romenelli 1988) . 



12 
An interesting observation made during these in vitro 
transcription studies was that, in all species tested, both 
light and heavy strand promoters have the ability to direct 
in vitro transcription from either DNA strand from a common 
initiation point (Bogenhagen and Yoza 1986; Chang et al. 
1986) . Contrary to what might be expected, the DNA 
seguences surrounding some of these bidirectional initiation 
sites are not inverted repeats. This bidirectional nature 
of mitochondrial promoters has thus far been seen only in 
the human in vivo . 

Mitochondrial Transcription Factors 

Further in vitro studies in human, mouse, and Xenopus 
systems have uncovered the existence of a mitochondrial 
transcription factor that can be separated from the RNA 
polymerase activity (Fisher and Clayton 1985; Bogenhagen and 
Insdorf 1988; Fisher et al. 1989). The presence of these 
proteins is reguired for efficient in vitro recognition of 
the D-loop transcriptional starts sites by their respective 
RNA polymerase activities. The human mitochondrial 
transcription factor 1 (mitTFl (Fisher and Clayton 1988)) 
and mouse factors have been shown to bind specifically to 
limited DNA regions (approximately at positions -10 to -40) 
upstream of their respective light strand transcriptional 
start sites (Fisher et al. 1987; Fisher and Clayton 1988; 
Fisher et al. 1989) . As mentioned above, mutagenesis of the 






13 

human mitTF binding sequence at the light strand promoter 
has been shown to significantly reduce in vitro 
transcription at the associated transcription initiation 
site (Chang and Clayton 1984; Fisher et al. 1987) . DNA 
binding by mitTFl at sequences upstream from the human heavy 
strand RNA start site is less specific (Fisher et al. 1987; 
Fisher and Clayton 1989) . DNA binding at this region by 
mitTFl is only detectable in DNAse 1 protection assays when 
an excess of the factor is present, with the resultant 
footprint still being faint and without defined boundaries. 
Analogous specific binding by the mouse mitTF to its 
homologous heavy strand promoter region has not been 
reported. 

Even though human and mouse transcription factors have 
been shown to recognize and bind to defined DNA sequences, 
the nature of this binding is not strongly sequence specific 
since it is readily diminished in vitro by the addition of 
non-specific competitor DNA (Fisher and Clayton 1988; Fisher 
et al. 1989) . The weak specificity of these proteins is 
further demonstrated by heterologous assays in which the 
ability of human and mouse factors to recognize and bind to 
analogous regions upstream from both the human and mouse 
light strand promoter sequences has been demonstrated, even 
though these regions contain no detectable sequence 
similarity (Fisher et al. 1989). The two factors can also 
functionally substitute for each other in heterologous in 



14 
vitro transcription experiments, each with the ability to 
stimulate accurate transcription by the heterologous RNA 
polymerase (Fisher et al 1989) . More recently mitTFl has 
been cloned and sequenced (Parisi and Clayton 1991) . Amino 
acid similarities to proposed DNA-binding domains of 
different types of high mobility group (HMG) proteins were 
noted (Parisi and Clayton 1991) . HMG proteins have been 
shown to interact with DNA in a variety of processes 
including chromatin formation and transcriptional 
activation, but in no instance has a strong DNA sequence 
preference been demonstrated. 

Why Study Artiodactyl Mitochondrial Transcription? 

In an effort to understand the functionality of D-loop 
DNA sequences in the LSP/HSP regulatory regions in the face 
of the significant sequence variation that occurs between 
divergent mammalian species, we have studied mitochondrial 
transcription initiation in a series of more closely related 
species from the Artiodactyl order. Our investigations have 
focused on transcription at the LSP since RNA synthesis at 
this site may also be relevant to DNA replication. We 
initially mapped L-strand RNA start sites in two of the most 
divergent Artiodactyl species (cow and pig) and used these 
sites to align and compare D-loop sequences of other 
Artiodactyls. Using this alignment we were able to identify 
an 8 bp sequence strictly conserved at the putative L-strand 



15 

RNA start sites of all Artiodactyls. Blocks of sequence 
homology were also noted further upstream whose evolutionary 
conservation suggests their potential role as 
transcriptional regulatory domains. Curiously, we found 
that the porcine sequence contained a 20 bp insertion 
relative to the other Artiodactyls at a position about 10 bp 
upstream of the L-strand start site: the middle of a region 
shown in other mammalian species to be important for mitTF 
binding and for regulating efficiency of L-strand 
transcription . 

In order to test the functional relevance of the 
conserved Artiodactyl domains, we developed in vitro 
transcription systems from the mitochondria of both pig and 
cow. Previous to these investigations all mitochondrial RNA 
polymerase preparations from other mammalian species were 
isolated from tumor cell lines grown in culture (Walberg and 
Clayton 1983; Chang and Clayton 1986a). In an attempt to 
achieve an accurate representation of in vivo transcription, 
we report here the first isolation of mitochondrial 
transcription systems from adult tissue. The protein 
extracts, isolated from brain tissue of both pig and cow, 
were capable of accurate initiation of RNA synthesis on 
homologous cloned DNA templates. Heterologous transcription 
experiments showed, somewhat surprisingly, that although 
significant sequence differences exist between the porcine 
and bovine light strand promoter sequences, the two RNA 



16 
polymerase preparations were functionally interchangeable: 
both being able to accurately transcribe on either the 
porcine or bovine template. 

Extending our transcription studies, we were able to 
identify and partially separate a bovine mitochondrial 
transcription factor activity from the bovine RNA polymerase 
activity. Using both band shift and DNAse 1 protection 
assays, sites of specific protein :DNA interaction upstream 
from both the bovine light and heavy strand start sites were 
detected. In heterologous assays, the bovine transcription 
factor was able to recognize and bind specifically to a wide 
range of sequences, including regions upstream from the 
light and heavy strand promoters of the human and mouse, and 
the light strand promoter of the pig. A sequence alignment 
of promoter regions bound specifically by the factor 
suggests that only rather general DNA sequence motifs are 
required for transcription factor recognition. Sequence 
comparisons with mitochondrial regions outside the bovine D- 
loop show several potential recognition sites for 
transcription factor binding. Further studies using a 
synthetic Artiodactyl LSP template, engineered with the 
conserved sequence domains in mobile cassettes, suggests 
that when high levels of mitTF activity are present, 
specific protein: DNA interaction by the factor is not 
required for accurate transcription initiation in vitro . 
These results support the hypothesis that transcription 



17 

occurs in regions of the mitochondrial genome other than the 
D-loop. They also lead to a unified picture of 
mitochondrial transcription in which the level of mitTF 
serves to regulate preferential LSP directed RNA initiation. 












CHAPTER 2 
MATERIALS AND METHODS 



Isolation of Mitochondria from Brain Tissue 

Usually 2 to 3 brains from adult animals (obtained 
immediately after slaughter from the either the Meat 
Sciences Department at the University of Florida (cows) or 
Nettle's Sausage Company in Lake City, Florida (pigs)) were 
processed together during a single preparation of 
mitochondria. In order to maintain the integrity of the 
mitochondrial proteins, all solutions were pre-chilled on 
ice and all procedures were done at 4°C and on ice when 
possible. Approximately 600 g of brain tissue and 500 mis 
of an ice cold solution of 50 mM Tris-HCl, pH 7.5, 0.21 M 
mannitol, 70 mM sucrose, and 3 mM CaCl 2 (MSB-Ca++) per 
brain, were placed in a blender and homogenized on medium 
speed until the tissue was mostly liquefied (about 10 to 20 
seconds on medium speed) . The tissue was further 
homogenized by 3 strokes in a glass homogenizer with a 
tightly fitting motor-driven pestle. The total volume of 
the homogenate was measured, and 0.5 M EDTA was added to a 
final concentration of 20 mM. The homogenate was spun at 
3,000 rpm in a Beckman J-10 rotor for 10 min to pellet 
nuclei and cellular debris. The reddish supernatant was 

18 



19 
collected and respun as above. The supernatant was again 
collected and centrifuged at 13,000 rpm in a Beckman J-17 
rotor for 20 min to pellet the mitochondria. The 
supernatant was discarded and the mitochondrial pellet 
washed in 5 volumes of 50 mM Tris-HCl pH 7.5, 0.21 M 
mannitol, 70 mM sucrose, and 10 mM EDTA (MSB EDTA) and 
repelleted as before. The pellet was then resuspended in a 
2 0% sucrose solution containing 50 mM Tris-HCl pH 7.5, and 
10 mM EDTA, and the mitochondria banded in a sucrose density 
step gradient between 1.0 M and 1.5 M sucrose in 10 mM Tris- 
HCl pH 7.5, and 5 mM EDTA by centrifugation at 22,000 rpm 
for 20 min in a Beckman SW-27 rotor. The purified 
mitochondria were collected and diluted 2-fold with MSB 
EDTA, aliguotted into 6 to 8 portions and repelleted in a 
Beckman JA-2 rotor at 15,000 rpm for 10 min at 4°C. The 
supernatants were discarded and the pelleted mitochondria 
were either used immediately or stored at -70°C. 
Mitochondria prepared and stored as above lasted 3 months or 
more without significant loss of protein activty. 

Isolation of Mitochondrial Nucleic Acids 

A 1 g mitochondrial pellet was suspended in 25 ml of an 
ice-cold solution containing 3 M LiCl, 6 M urea, and 0.2% 
SDS and homogenized using 10 strokes of a tightly fitting 
dounce homogenizer. Nucleic acids were then precipitated by 
incubation of the homogenate overnight on ice and pelleted 



20 
by centr if ligation for 30 min at 4°C and 4500 rpm in a 
Beckman J-20 rotor. The pellet was resuspended in 10 mM 
Tris-HCl pH 8.0, 0.5% SDS, and 1 mM EDTA and sequentially 
extracted with equal volumes of phenol, phenol: chloroform 
(1:1, v/v) and chloroform. Nucleic acids were again 
precipitated by the addition of ammonium acetate to 0.4 M 
and 3 volumes of ethanol. Mitochondrial nucleic acids 
prepared as above and stored in ethanol at -20°C were stable 



for at least 1 year, 






Labeling of Probe DNAs 

DNA fragments were labeled at their 5' ends using T4 
polynucleotide kinase. Typically about 20 ug of plasmid was 
linearized with an appropriate restriction enzyme. After 
complete digestion (as determined by electrophoresis of an 
aliquot of the reaction on a 1% agarose gel) , 1 unit of calf 
intestinal phosphatase (Boeringer Mannheim) was added to the 
reaction mix and incubated at 37°C for 20 min, followed by 
the addition of a second unit of phosphatase and another 20 
min incubation. The phosphatase was inactivated by 
incubation at 65°C for 45 min in the presence of 10 mM 
nitriloacetic acid. Sodium acetate (pH 5.2) was then added 
to a final concentration of 0.3 M and the reaction mix 
extracted twice with equal volumes of 

phenol: chloroform : isoamyl alcohol (1:1:25, W/V) . After the 
addition of 3 volumes of 95% ethanol, the mixture was placed 



21 
on ice for 10 min and then spun at 15,000 rpm in a 
microcentrifuge to pellet the DNA. The pellet was 
resuspended in 100 ul of 10 mM Tris-HCl pH 8.0, and 1 mM 
EDTA (TE) and used immediately or stored at -20°C. A 5 ul 
aliquot (about lug) of the digested DNA was placed in a 20 
ul reaction volume containing 50 mM Tris-HCl pH 8.3, 6.0 mM 
MgCl 2 , 40 mM KC1, 1.2 uM gamma- 32 P ATP (ICN Biochemicals) , 
and 1 unit of T4 polynucleotide kinase (U.S. Biochemical) 
and incubated for 1 hr at 37°C. After incubation the 
nucleic acids were phenol: chloroform extracted and ethanol 
precipitated as above. 

For probe DNAs uniquely labeled at one 5' end, the 
pellet was suspended in an appropriate reaction buffer (as 
recommended by the restriction enzyme supplier) and digested 
with a second restriction enzyme to generate 2 labeled DNA 
fragments. The labeled fragments were then resolved on 5% 
non-denaturing acrylamide gels and visualized by exposure to 
X-ray film (Kodak, XAR-5) . The desired fragments were cut 
out of the gel and incubated overnight at 37°C in 10 volumes 
of TE. The gel slice was then removed from the solution 
containing the eluted DNAs. Sodium acetate was added to a 
final concentration of 0.3 M and the labeled DNAs 
precipitated by the addition of 3 volumes of ethanol. 

For labeling primer DNA oligonucleotides, about 15 ng 
of primer was suspended in a kinase reaction mix (as above) 
and incubated for 1 hr at 37°C. Following incubation, 



22 

unincorporated labeled ATP was removed from the reaction mix 
by two successive dilutions with 1 ml of water and 
centrifugations in a Centricon 10 (Amicon) for 1 hr at 6,500 
rpm. 

DNA 3 * ends were labeled using the Klenow fragment of 
DNA polymerase I (BRL) . Similar to the kinase reaction 
about 1 ug of linearized plasmid DNA (restriction enzyme 
digested to create 3' recessed ends) was incubated for 1 hr 
at 37°C in a 50 ul reaction volume containing 50 mM Tris-HCl 
pH 7.5, 3 mM MgCl 2 , 1 mM dithiothreitol, and a mixture of 1 
uM of each of 2 dNTPs labeled at the alpha position with 32 P 
at 3000 Ci/mmole (the precise dNTPs depended on the DNA 
seguence of the restriction site to be filled in by the 
polymerase) and the remaining dNTPs at 0.1 mM. After 
incubation for 15 min the extension reaction was chased to 
completion by the addition of unlabled dNTPs at 0.1 mM. 
Following the hour long incubation, the reaction was 
phenol: chloroform extracted and ethanol precipitated. 
Similar to the kinase reaction above, for probes uniguely 
labeled at the 3' end, DNAs were digested with a second 
restriction enzyme and isolated after electrophoresis on 5% 
acrylamide gels. 

DNAse 1 and RNAse Tl Digestion of Nucleic Acids 

Nucleic acids were suspended in a 50 ul reaction volume 
containing 10 mM Tris, pH 8.0, 5 mM MgCl 2 , and 10 Mm CaCl 2 . 



23 

For DNAse 1 digestion, aliquots between 0.012 and 12 ug of 
DNAse 1 were added and the mixture incubated for 1 hr at 
37°C. For RNAse digestion, 500 units of RNAse Tl were added 
and the mixture incubated at 37°C for 1 hr. After DNAse or 
RNAse digestion, 10 ug of carrier tRNA, sodium acetate 
acetate (to 0.3 M) and EDTA (to 10 mM) were added. The 
reaction mixture was then extracted with phenol, 
phenol: chloroform, and chloroform and ethanol precipitated. 

SI Nuclease Protection Assays 

The SI probe was made by 5 'end labeling a Bam HI digest 
of a plasmid containing a 592 bp Bam HI-HincII insert 
(corresponding to bases 16202 to 457 on the bovine 
mitochondrial sequence) directionally ligated into the 
Bluescript (Stratagene) polylinker region. The labeled DNA 
was then digested with Hinc II and electrophoresed on a 5% 
non-denaturing acrylamide gel. After visualization with 
autoradiography, the 592 bp probe was excised from the gel 
and the gel slice incubated overnight at 37°C in 0.5 ml of 
10 mM Tris pH 8.0, 1 mM EDTA. Sodium acetate was added to a 
final concentration of 0.3 M and nucleic acids ethanol 
precipitated. The labeled probe (typically about 100,000 
cpm) and 10 ug of mitochondrial nucleic acid either 
previously untreated or DNAse I and/ or RNAse Tl digested 
(see above) were mixed in 50 ul of TE and ethanol 
precipitated. The nucleic acid pellet was then resuspended 



24 
in a 30 ul volume of 20 mM Tris-HCl pH 7.4, 1 mM EDTA, 0.4 M 
NaCl, and 80% formamide and heat denatured at 75°C for 10 
min. The mixture was immediately placed in a water bath at 
50°C and incubated for 3 hr to allow annealing of the 
labeled probe to the target sequences. Afterward, 0.3 ml of 
an ice-cold mixture containing 0.28 M NaCl, 0.05 M sodium 
acetate pH 5.2, 4.5 mM ZnS0 4 , 20 ug/ml heat denatured 
herring sperm DNA, and 300 units/ml of SI nuclease (BRL) was 
added and the reaction incubated for 1 hr at 37°C to digest 
single stranded nucleic acids. The reaction was then 
phenol: chloroform extracted and then sodium acetate (to 0.3 
M) and 3 volumes of ethanol were added. The ethanol mixture 
was placed on ice for 10 min and then spun at 15,000 rpm in 
a microcentrifuge for 15 min to pellet the nucleic acids. 
The pellet was resuspended in 4 ul of 95% formamide, 0.1% 
bromophenol blue, 0.1% xylene cyanol, heated to 90°C for 2 
min, and electrophoresed on 6% polyacrylamide, 7 M urea 
gels. After electrophoresis the gels were dried and labeled 
species were visualized by autoradiograpy with X-ray film. 

Primer Extension Assays 

Synthetic oligonucleotide primers were labeled at the 
5' end using gamma 32 P-labeled ATP and T4 polynucleotide 
kinase (see above) . The labeled primers were mixed with 
nucleic acids (previously either untreated or after 
digestion, as above, with DNAse 1 and/or RNAse Tl) from 



25 
mitochondria or in vitro transcription reactions, in 15 ul 
of water, heated to 70°C for 10 min, and immediately placed 
on ice for 5 min to allow annealing of the labeled primer. 
A 15 ul reaction volume containing 50 mM Tris-HCl pH 8.3, 75 
mM KC1, 3 mM MgCl 2 , 10 mM dithiothreitol, 1 mM each of dATP, 
dGTP, dCTP, dTTP and 100 units of M-MLV reverse 
transcriptase (BRL) was then added to the annealing mixture, 
and the reaction incubated at 37°C for 2 hrs. The nucleic 
acids were then ethanol precipitated and electrophoresed on 
6% acrylamide 7.5 M urea denaturing gels. 

RNA Sequencing 

Direct dideoxynucleotide seguencing of mitochondrial 
RNA was performed in a manner similar to that of Geliebter 

(1987) . Ten ug of total porcine mitochondrial RNA was mixed 
with 10 ng of a 5 1 32 P end-labeled oligonucleotide 

(corresponding to bases 1059 to 1076 on the procine 
seguence) in a 12 ul solution containing 260 mM KC1 and 10 
mM Tris, pH 8.3. The mixture was heated at 80°C for 3 
minutes, and the primer annealed to the RNA by incubation of 
the sample at 52°C for 45 minutes. Two ul of the primer: RNA 
solution was then aliguotted into 4 tubes (for each base 
specific seguencing reaction) containing 3.3 ul of reverse 
transcription buffer (50 mM Tris , pH 8.3, 6 mM MgCl 2 , 40mM 
KC1, 0.4 mM dATP, 0.4 mM dCTP, . 8 mM dGTP, and 0.4 mM 
dTTP) , 2 units of AMV reverse transcriptase and 1 ul of 



26 
either 2 mM ddATP, 1 mM ddCTP, 1 mM ddGTP, or 1 mM ddTTP. 
The mixture was then incubated at 50°C for 45 minutes and 
the reaction terminated by addition of formamide-dye. The 
samples were denatured by boiling 3 min and electrophoresed 
on a denaturing 6% polyacrylamide gel. 

Cloning of DNA Fragments 

pR16 . The 5' half of the bovine D-loop (bases 16202- 
457) was isolated by digestion of the pCMR plasmid with Bam 
HI and Hinc II and electrophoresis of the digestion products 
on a 1% agarose gel. A slice was then made in the gel in 
front of the DNA fragment of the correct size, where a piece 
of NA45 membrane (Schleicher and Schuell) was inserted. The 
DNA was electrophoresed onto the membrane, which was then 
removed and washed in TE buffer. Bound DNA was then eluted 
by heating at 55°C for 45 min in TE plus 1 M NaCl, followed 
by ethanol precipitation. The DNA was pelleted and 
resuspended in TE buffer. An aliquot containing 0.5 ug of 
the Bluescript vector (Stratagene) , previously digested with 
Bam HI and Hinc 11 (as per enzyme supplier's conditions) was 
incubated with 0.5 ug of the isolated DNA fragment in a 20 
ul reaction mix containing 66 mM Tris-HCl pH 7.6, 6 . 6 mM 
MgCl 2 , 10 mM DTT, 66 uM ATP and 1 unit T4 DNA ligase (BRL) 
for 1 hr at 37°C. The ligation mixture was then diluted 5- 
fold with water and 1 ul used to transform competent 






27 
DH5-alpha cells (BRL) . A diagram of the cloned fragment is 
shown in Figure 3-1. 

pALSP . Two complementary DNA oligonucleotides 
(CTTAAATACACCACCACTTTTAACA) were synthesized by the ICBR DNA 
Synthesis core facility at the University of Florida. 
Aliguots of 0.25 ug of each DNA strand were mixed together 
in a 20 ul volume of water. To anneal the two DNA strands, 
the mixture was heated to 65°C for 10 min and then slow 
cooled over 45 min to room temperature. After annealing, a 
10 ul aliguot of the DNAs was mixed with an aliquot 
containing 1 ug of the Bluescript vector (previously 
digested with Eco Rl and Sal 1 according to manufacturer's 
specifications) in a 20 ul ligation reaction mixture similar 
to above. A second abberant plasmid, pRRALSP, was a by- 
product of the above procedure, evidently the result of 
ligation of the annealed DNA oligonucleotides to a molecule 
of the Bluescript vector which had only been digested at the 
Eco Rl restriction enzyme site. 

Subclones of the pALSP and pRRALSP DNAs were made by 
digestion of either plasmid at two restriction enzyme sites 
(pPAB: pRRALSP digested with Bgl II and Eco RV; pPC+D: pALSP 
digested with Bgl II and Cla 1; pPA: pRRLSP digested with 
Hind III; pP: pRRALSP digested with Cla I; pABC+D: digested 
with Bam HI and Cla 1; pC+D: digested with Bam HI and Bgl 
II) , followed by the addition of dATP, dGTP, dCTP, and DTTP 
and 2 units of the Klenow fragment of DNA polymerase 1, to 



28 
fill in the staggered DNA ends of the restriction digest. 
After ethanol precipitation a 1 ug aliguot was incubated in 
a ligation mixture, for use in subsequent transformations. 
For pPA and pP, since digestion with one enzyme resulted in 
two compatible DNA ends, the klenow DNA polymerase step was 
omitted. 

Fractionation of Mitochondrial Proteins 

Typically, 1 g of pelleted mitochondria was thawed on 
ice and resuspended in 6 ml of 2X column buffer (IX column 
buffer is 10 mM Tris-HCl pH 8.0, 10 mM MgCl 2 , 1 mM EDTA, 1 
mM dithiothreitol, 1 mM phenylmethylsulfonylf luoride, 2 mM 
benzamide HC1, and 7.5% glycerol) at 0°C. The volume was 
adjusted to 8.4 ml with ice cold water. After adding 1.2 ml 
of 20% triton-X 100, the mitochondria were homogenized by 10 
strokes in a glass homogenizer. The homogenization was 
repeated after adding 3.6 ml of 4 M KCl. The lysate was 
then cleared by centrifugation in a Beckman Ti-50 rotor for 
30 min at 40,000 rpm at 4°C. The amber colored lysate was 
diluted about 4-fold with column buffer to adjust the KCl 
concentration to 0.2 M as determined by conductivity. The 
diluted mixture was then loaded onto a 60 ml heparin- 
Sepharose column at a flow rate of 0.5 column volumes/hr. 
The column was then washed with 3 column volumes of column 
buffer at 0.2 M KCl, until all unbound proteins had been 
removed as determined by O.D. at 280 nm of the eluate. 



29 

Typically, greater than 95% of the lysate protein was 
removed by this washing. Bound proteins were then eluted 
from the column using a step gradient with column buffer at 
0.6 M KCL at a flow rate of 1 column volume/hr. Fractions 
of 3 ml were collected and assayed for KCl concentration by 
conductivity and RNA polymerase activity. Active fractions 
(usually all fractions containing detectable proteins) were 
then pooled and either dialyzed in column buffer or diluted 
with the same to reduce the KCl concentration below 0.05 M. 
The pooled eluate was then loaded onto a 30 ml DEAE-sephacel 
column (previously eguilibrated with column buffer at 0.05 M 
KCL) at a flow rate of 1 column volume/hr. Similar to 
above, the column was washed with column buffer at 0.05 M 
KCL until all unbound proteins had been removed. Bound 
proteins were eluted using either a step gradient of column 
buffer at 0.5 M KCl or a linear gradient between 0.05 and 
0.5 M KCl, at a flow rate of 1 column volume/hr. Fractions 
of 2 ml were collected, assayed for KCL concentration by 
conductivity, and dialyzed individually into 10 mM Tris pH 
8.0, 10 mM MgCl 2 , . 5 mM KCl, 0.1 mM EDTA, 1 mM 
dithiothreitol, 1 mM phenylmethlysulfonylf louride, 1 mM 
benzamide HC1, and 50% glycerol at 4°C. Dialyzed fractions 
were assayed for RNA polymerase activity and stored at - 
20°C. 



30 
RNA Polymerase Assays 

Non-specific RNA polymerase activity . Aliquots of 10 
ul from each column fraction were incubated in 100 ul 
reaction mixtures with 10 mM Tris. pH 8.0, 10 mM MgCl 2 , 100 
ug/ml bovine serum albumin, 1 mM dithiothreitol, 500 uM ATP, 
150 uM each of CTP and GTP, and 150 ug/ml of heat-denatured 
salmon sperm DNA. UTP at 0.005 uM was included as alpha- 3 2P 
UTP, 3 000 Ci/mmol. The reactions were incubated at 37°C for 
20 min. Reactions were precipitated by the addition of 15 
volumes of 7.5% TCA, followed by incubation at 0°C for 20 
min. Reaction products were collected onto nitrocellulose 
filters, washed with 50 ml of 7.5% TCA, and assayed in 
ScintiVerse E (Fisher) in a scintillation counter. RNA 
polymerase activity was measured by the ability of each 
fraction to incorporate the radio-labeled ribonucleoside 
triphosphates into acid-precipitable counts. 

Specific Transcription Initiation . Cloned plasmid 
DNAs, either supercoiled or linearized (as specified in the 
figure legends) at 2 ug/ml were incubated for 3 min at 
37°C with 4-8 ul of the mitochondrial protein fractions in a 
50 ul reaction volume containing 10 mM Tris. pH 8.0, 5 mM 
MgCl 2 (except where indicated) , 100 ug/ml bovine serum 
albumin, 1 mM dithiothreitol, 500 uM ATP, and 150 uM each 
of CTP, GTP, and UTP. For run-off transcription assays the 
unlabeled UTP was replaced by alpha- 32 P UTP (3000 Ci/mmol) 
at 0.005 uM. Reactions were terminated by the addition of 



31 

EDTA (to 10 mMM) and sodium acetate (to 0.3 M) . Products of 
the transcription reactions were extracted with phenol and 
phenol: chloroform, and ethanol precipitated. Reaction 
products from the run-off assays were then electrophoresed 
on 6% acrylamide, 7.5 M urea gels. Unlabeled transcription 
reactions were then either untreated, or digested with DNAse 
I and/or RNAse TI for further use in primer extension 
assays. 

Nuclease Assays 

DNA nuclease assay . A cloned Bam HI-Hpa I DNA fragment 
corresponding to bases 16202-457 on the bovine mitochondrial 
genome was 32 P 5' end labeled at the Bam HI site on either 
the coding or non-coding strand. 0.25 ug of each 
radiolabelled fragment was then incubated in a 50 ul 
transcription buffer (10 mM Tris pH 8.0, 10 mM MgCl 2 , 100 
ug/ml bovine serum albumin, 1 mM dithiothreitol, 500 uM ATP, 
150 uM each of CTP and GTP, where 32 P UTP was replaced with 
unlabeled UTP at a concentration of 150 uM) with 5-10 ul of 
a column fraction at 37°C for 3 min. The reactions were 
stopped by the addition of EDTA to 10 mM. The reactions 
products were then ethanol precipitated and electrophoresed 
on 6% acrylamide, 8 M urea sequencing gels. 

RNA nuclease assay . Reactions to detect contaminating 
RNA nucleases in column fractions were performed essentially 
in the same manner as the DNAse assays above except that 32 P 



32 
labeled RNA transcripts (generated using the cloned mitDNA 
from above, linearized at the Bam HI site, as template for 
T3 RNA polymerase in run-off transcription reactions) were 
substituted for the labeled DNA. 

DNA Binding Assays 

DEAE-sephacel purified DNA binding proteins were 
incubated in a 15 ul reaction volume for 20 min at room 
temperature with 0.5 -1 ng of 32 P labeled DNA fragments 
under conditions similar to those used in transcription 
assays except that glycerol was added to a final 
concentration of 10% and nucleoside triphosphates were 
omitted. After eguilibration, the reaction mixtures were 
electrophoresed over 5% acrylamide gels in 10 mM Tris-HCl pH 
8.0, 1 mM EDTA. After electrophoresis, the gels were dried 
and exposed to X-ray film. 

For DNAse 1 protection, binding proteins were incubated 
at room temperature with DNA fragments as above, but were 
subjected to limited digestion by the addition of 15 ul of 
20 mM Tris-HCl pH 8.0, 5 mM MgCl2, 10 mM CaC12 and 0.5-2.0 
ug/ml DNAse I (BRL) for 20 sec. Following addition of 150 
ul of stop buffer (0.33 M sodium acetate, 11 mM EDTA, 1% SDS 
and 60 ug/ml tRNA) , samples were extracted once with 
phenol: chloroform. Nucleic acids were recovered by ethanol 
precipitation and electrophoresed on 6% acrylamide 7.5 M 
urea gels. 



CHAPTER 3 
SITES OF ARTIODACTYL L-STRAND RNA INITIATION 



Introduction 
Transcription of vertebrate mitochondrial DNA is 
believed to be regulated from only two promoter regions 
(light and heavy strand promoters: LSP and HSP) , each 
responsible for directing RNA synthesis of the entire genome 
from its respective DNA strand (Aloni and Attardi 1971a; 
Murphy et al. 1975; Montoya et al. 1982; reviewed in Clayton 
1984) . In all vertebrate species studied to date (human, 
mouse, and Xenopus ) both promoters map to the region of the 
D-loop between the origin of heavy strand replication (OH) 
and the phenylalanine tRNA gene (Murphy et al. 1975; Montoya 
et al. 1982; Yoza and Bogenhagen 1984; Chang and Clayton 
1986a) . The start sites of light strand transcription for 
these species lie approximately 150-300 bases upstream of 
the OH with conserved seguence blocks (CSBs) -1, -2, and -3 
interspersed between. Transcription from the LSP region is 
thought to prime mitochondrial DNA replication through the 
use of CSB regions as recognition signals for the transition 
from RNA to DNA synthesis. Presumably CSBs are recognized 
by the RNA component of a site specific mitochondrial RNAse 

33 



34 
activity which cleaves L-strand transcripts to create a 
primer RNA 3 * -OH (Chang and Clayton 1987). 

Experiments using in vitro transcription systems have 
shown apparent differences between the size of mammalian and 
amphibian mitochondrial transcriptional regulatory regions. 
In vitro , mammalian mitochondrial light strand promoters 
consist of a core seguence (from about +16 to -28) reguired 
for accurate initiation and regions further upstream (up to 
about -56) which are important in regulating transcription 
efficiency (Chang and Clayton 1984; Chang and Clayton 1986b; 
Chang and Clayton 1986c) . Similar to the yeast system 
(Biswas et al. 1985) mitochondrial transcription in Xenopus 
seems only to reguire a small seguence (from about +9 to -9) 
for wild type activity in vitro (Bogenhagen and Romenelli 
1988) . Despite these apparent differences in promoter size, 
an unusual characteristic of all vertebrate mitochondrial 
promoter seguences tested is their ability to direct RNA 
synthesis bidirectionally on both template DNA strands from 
a single initiation site (Bogenhagen and Yoza 1986; Chang et 
al. 1986a) . The bidirectional nature of these regulatory 
elements has been documented primarily using in vitro 
transcription experiments but has also been identified in 
vivo in human cells (Chang et al 1986a) . 

Direct seguence comparison of promoter elements from 
these species demonstrates that while overall D-loop 
organization is conserved, the DNA seguences immediately 



35 
surrounding the light strand RNA start sites are very 
heterogenous (Chang and Clayton 1986a) . This fact has 
prohibited a dissection of promoter structure by the 
criterion of sequence conservation, usually a powerful 
method to identify critical DNA domains. Our investigations 
compare mitochondrial promoter sequences in a series of more 
closely related species, attempting initially to infer DNA 
sequences or sequence blocks relevant to transcription by 
their evolutionary conservation. We have begun these 
studies by attempting to identify potential regulatory 
sequences in the displacement loop region of four species 
within the Artiodactyl order (cow, water buffalo, giraffe, 
and pig) . A comparison of the DNA sequences near putative 
light strand transcriptional initiation sites in these 
species identifies a series of common sequence domains 
identically located relative to experimentally determined 
light strand RNA 5' ends for cow and pig. The evolutionary 
conservation of these domains suggests their possible role 
in regulating mitochondrial transcription. 

Results 

Identification of the Bovine Light Strand Transcriptional 
Initiation Site 

From the investigations of MacKay (Ph.D. Thesis, 

University of Florida) and Tanhauser (Ph.D. Thesis, 

University of Florida) a series of cloned mitochondrial D- 

loop sequences were available to our laboratory. These DNA 



36 
clones were derived from a variety of exotic as well as 
domestic species of the Artiodactyl order. Since bovine and 
porcine tissues were readily obtainable from the Department 
of Animal Sciences at the University of Florida, our studies 
originally focused on these species. 

To determine the transcription initiation sites from 
the bovine light strand promoter we initially constructed a 
5* P labeled, single stranded DNA probe complementary to 
the mitochondrial DNA of the bovine heavy strand and large 
enough to span the entire 5 ■ region of the D-loop 
(corresponding to bases 16202 to 457 on the bovine seguence, 
diagrammed in Figure 3-1, Panel B) . When used in SI 
protection analyses after hybridization to total bovine 
mitochondrial nucleic acid, a series of 5' ends were 
detected ranging in size from about 200 to 245 nucleotides 
(Figure 3-1, lane 3) ; the most prominent species mapped 
approximately to positions 240-245, with species of lesser 
intensity at 228-232 and 203-204 nucleotides. Pretreatment 
of nucleic acids with RNAse Tl (lane 2) removed the nucleic 
acid responsible for the major band, while some of the bands 
of lesser intensity were still present, indicating that the 
major band was composed primarily of RNA 5' ends while the 
minor bands have at least some DNA component. Upon 
pretreatment with DNAse 1 (lane 1) , the nucleic acids 
responsible for the major band were totally resistant, and 
components of the minor bands were also partially resistant. 






Figure 3-1. S-l nuclease protection of bovine mitochondrial 
L-strand RNA . Panel A. Mapping of the bovine L-strand RNA 
5 1 end. A DNA probe 5' 32 P radiolabeled at the Bam HI site, 
corresponding to bases 16202-457 of the bovine mitDNA 
sequence (shown in Panel B) , was annealed to total 
mitochondrial nucleic acid (lane 3) , mitochondrial nucleic 
acids pretreated with RNAse Tl (lane 2) , or pretreated with 
DNAse 1 (lane 1) and incubated with S-l nuclease as 
described in materials and methods. Lanes C/T and A/G 
represent Maxam and Gilbert (1980) sequencing reactions of 
the probe DNA, used for size reference. Approximate 
nucleotide positions are shown in the margin. Panel B. 
Diagram of pR16. A 592 bp Bam Hl-Hinc II fragment of the 
bovine mitochondrial genome was directionally cloned into 
the polylinker of the Bluescript vector (indicated by heavy 
lines) . This section of the bovine mitochondrial genome 
contains the entire 5' region of the D-loop, including the 
origin of heavy strand replication (OH) , conserved sequence 
blocks 1 and 2+3 (CSB-1 and CSB-2+3) (see text), the 
phenylalanine tRNA gene (tRNAphe) and the first 27 bp of the 
12S rRNA gene. The T3 promoter region contained within the 
Bluescript vector is shown by the dark box. 



38 



B. 



A 



C/T A/G 3 




Bam HI 



16202 



OH 



Hlnc II -. 
tRNAphe 

i i . ; 



CSB -1 -2+3 



12S rRNA 



V 



457 
































39 

These findings indicated the existence of a single major 
light strand RNA 5' end at position 240-245 with species 
downstream representing a mixture of RNA and DNA 5' ends. 
Independent investigations by King and Low (1987) , using 
mitochondrial RNA isolated from bovine tissue culture cells 
also identified the major light strand 5' RNA end near this 
region. 

The absence of available restriction sites in the 5' 
region of the bovine D-loop limited our ability to construct 
probes of a size necessary for high resolution mapping using 
SI protection analysis. To overcome this problem, we 
synthesized a DNA oligonucleotide corresponding to bases 
136-156 (approximately 100 bases downstream from the 5' ends 
identified above) for use in primer extension experiments. 
In order to further test the observation that some of the SI 
protected 5 1 ends had both RNA and DNA components, prior to 
annealing and extending the 5 ' 32 P end labeled primer with 
AMV reverse transcriptase, we pretreated 10 ug aliquots of 
mitochondrial nucleic acids with various concentrations of 
DNAse I (Figure 3-2, lanes 1-5). In parallel experiments we 
pretreated a set of aliquots with similar amounts of DNAse I 
as above and in addition added an excess of RNAse TI to each 
tube to remove all mitochondrial RNA (lanes 6-10) . Using 
this type of assay, any radiolabeled bands due to RNA should 
have a constant intensity in lanes 1-5, but be totally 
absent from lanes 6-10, while bands due to DNA 5' ends 



Figure 3-2. Precise mapping of the bovine mitochondrial L- 
strand RNA 5 ' end . Panel A. Primer extension of bovine 
mitochondrial L-strand RNA. Total mitochondrial nucleic 
acid was annealed to a DNA oligonucleotide (corresponding to 
bases 136-156 on the bovine sequence in Figure 3-5) , 
radiolabeled with 32 P at the 5' end, and primer extended 
with reverse transcriptase. The addition of ng (lanes 1 
and 6), 12 ng (lanes 2 and 7), 120 ng (lanes 3 and 8), 1.2 
ug (lanes 4 and 9), 12 ug (lanes 5 and 10) of DNAse 1 and 1 
ug of RNAse Tl (lanes 6-10) was used to treat the nucleic 
acids prior to annealing and primer extension. The 
corresponding nucleotide sequence of the bovine L-strand DNA 
is shown alongside. The bold arrows indicate the major 
transcription initiation sites and the direction of 
synthesis while the light arrows indicate minor RNA start 
sites. Asterisks denote labeled species of DNA origin. 
Panel B. Diagram of the bovine major L-strand RNA start 
site. The arrow labeled LSP indicates the relative position 
of the major site of L-strand transcription initiation 
(mapped above) within the bovine D-loop region and the 
direction of RNA synthesis. Approximate nucleotide 
positions are shown underneath. 






41 



12 3 4 5 6 



B 




Bam HI 



I 

16202 



CH 



LSP 



CSB -1 -2+3 



tRNAphe 



Hlnc II 



I2S rRNA 



=5^ 



42 
should gradually lose intensity equivalently in lanes 1-5 
and 6-10. As seen in Figure 3-2, this approach identified a 
major light strand transcription start site, a doublet of 
bands at positions 241 and 242 (lanes 1-5) . A faint duplex 
of RNA 5' ends were also seen 19 bp upstream at positions 
260 and 261. Present downstream were a series of less 
prominent 5 ■ ends that were a mixture of RNA and DNA 
(indicated by asterisks in the figure) . Since the origins 
of heavy strand replication in bovine mitochondria lie about 
50-130 bases downstream from this region (S. Mackay, Ph.D. 
thesis, University of Florida) , and since we have identified 
nuclease activities which digest nucleic acids in this 
region (Chapter 5) the RNA and DNA 5' ends identified 
immediately downstrem from position 241 are most likely due 
to nicks in the mitochondrial nucleic acids at these sites. 
The results of these primer extension analyses confirm and 
extend our original findings from SI protection analysis 
discussed above. 

Having identified a putative bovine light strand 
transcription initiation site we wanted to determine if RNA 
synthesis from this site was bidirectional. Detection of an 
RNA species of opposite polarity with a 5 • end near position 
240 would support the idea that the 5' end identified above 
is the actual light strand initiation site. Since this type 
of transcription would result in RNA species of opposite 
polarity we synthesized a DNA oligonucleotide (corresponding 



Figure 3-3. Primer extension analyses of bidirectional RNA 
from the bovine LSP . Panel A. Primer extension of H-strand 
transcripts initiating from the LSP region. A 5' P 
radiolabeled DNA oligonucleotide complementary to 
nucleotides 306-327 (on the bovine mitDNA) was annealed to 
total mitochondrial nucleic acids after pretreatment with 
ug (lanes 1 and 5), 12 ng (lanes 2 and 6), 120 ng (lanes 3 
and 7), and 1.2 ug (lanes 4 and 8) DNAse 1 and 1 ug RNAse Tl 
(lanes 5-8) . Oligonucleotide DNA: RNA hybrids were then 
primer extended and electrophoresed on sequencing gels 
alongside a 32 P labeled, Hpa II digest of pBR322 (lane M) . 
The arrow indicates the 5' end of the major H-strand RNA 
species transcribed from the bovine LSP region which map to 
positions 239 and 240 on the bovine sequence. Panel B. 
Diagram of bidirectional transcription initiating within the 
LSP region of the bovine D-loop. The heavy arrow indicates 
L-strand transcription and the direction of RNA synthesis, 
and the light arrow represents bidirectional H-strand 
transcription . 



A. 



M1 2345 678 



44 



B. 



♦ 






Bam Ml 



H 



1620? 



CH 



=\_ 



CSB -1 -2)3 



^ ISP 



tRNAphe 



Mine II 



12S rRNA 



7 



457 



45 
to nucleotides 329-3 07) complementary to the light strand 
DNA sequence for use in primer extension 

experiments. The oligonucleotide was 32 P end labeled at its 
5' end and annealed to aliquots of total mitochondrial 
nucleic acid similar to figure 3-2. This assay produced a 
single prominent DNAse I sensitive, RNAse Tl resistant site, 
a doublet which migrated at 87 and 88 nucleotides relative 
to the 5' end of the oligonucleotide (bases 239 and 240 on 
the bovine sequence (Figure 3-3, lanes 1-4)). The 
identification of a single prominent RNA species of opposite 
polarity (i.e. transcribed from the heavy stand DNA) whose 
5' end also maps to position 240 is independent evidence 
that we have mapped the bovine light strand start site. 

Identi fication of the Porcine Light Strand Transcriptional 
Initiation Site 

As mentioned briefly above, members of our laboratory 

had previously cloned and sequenced a 2.3 kb section of the 

porcine mitochondrial genome containing the proline tRNA 

gene, the D-loop region, the phenylalanine tRNA, and the 12S 

rRNA gene. These studies demonstrated extensive 

intraspecies sequence heterogeneity within the porcine D- 

loop. The majority of this heterogeneity was due to a 

series of ten base pair tandem repeats (monomer unit: 

ACGTGCGTAC) whose copy number varied from 14-29, located 

between the origin of heavy strand replication (OH) and 

CSB-2 (diagrammed in Figure 3-5, Panel C) . Attempts were 



46 
made to determine the porcine light strand transcriptional 
start site by SI protection analysis using a single stranded 
DNA probe that spanned the 5* end of the D-loop. Results 
from these experiments were difficult to interpret primarily 
due to seguence heterogeneity in the tandem repeat region 
between mitochondrial RNAs and the cloned DNA probe. Other 
problems of interpretation are discussed below. However, 
these studies gave an indication that light strand 
transcription began between bases 1100 and 1200 on the 
porcine D-loop seguence (S. MacKay, Ph.D. thesis, University 
of Florida) . To refine these initial findings we 
synthesized a DNA oligonucleotide complementary to the heavy 
strand DNA at bases 1054-1072 on the porcine seguence. This 
positioned the annealing site of the primer about 40 bases 
downstream of the region of interest, immediately upstream 
from CSBs-2 and -3 and the heterogenous repeated seguence 
region which we suspected had caused problems in 
interpreting the S-l experiments (diagrammed in Figure 3-4, 
Panel C) . When this oligonucleotide was used in primer 
extension experiments on total porcine mitochondrial nucleic 
acids, we detected a series of bands ranging in size from 75 
to 150 nucleotides (Figure 3-4, Panel A). Pretreatment with 
DNAse I revealed a single resistant species of about 85 
bases in length relative to the 5 ' end of the labeled primer 
(lane 2) , while pretreatment with RNAse Tl removed only the 
band at 85 nucleotides (lane 3) . Similar to primer 



Figure 3-4. Mapping of the porcine L-strand RNA 5' end . 
Panel A. Primer extension of porcine mitochondrial RNA. 
Total mitochondrial nucleic acid (lane 1) , after 
pretreatment with DNAse 1 (lane 2), or pretreatment with 
RNAse Tl (lane 3) was annealed to a 5 ' P end labeled 
primer (corresponding to nucleotides 1054-1072 on the 
porcine mitDNA sequence shown in Figure 3-5) and extended 
with reverse transcriptase. Lane 4 represents extension of 
only the labeled primer. Approximate DNA size standards are 
shown in the margin. Panel B. Run-off dideoxy sequencing 
of porcine L-strand mitochondrial RNA. As in Panel A, the 
labeled oligonucleotide was annealed to porcine 
mitochondrial nucleic acid, pretreated with DNAse 1, and 
extended using dideoxy sequencing reaction mixes. Lanes A, 
G, C, and T correspond to the dideoxy nucleotide added to 
each sequencing reaction. The arrow between Panels A and B 
corresponds to the 5' end of the porcine L-strand RNA. 
Panel C. Schematic diagram of the porcine D-loop. Open 
boxes represent the tandem repeats. The origin of heavy 
strand replication (arrow labeled OH), the 5' end of the 
porcine L-strand RNA (arrow labeled LSP) , CSBs-2 and -3 
(dark boxes) and the phenylalanine tRNA gene are also shown 
in the figure. Approximate nucleotide positions are shown 
at the bottom. 



A 



12 3 4 



1 60 — 



1 4 0— 



1 2 0_ 



1 0_ 



8 0— 



6 0— 



48 



.OM CSB 2 CSB 3 LSP IRHA PHE 

-D mniTiLumi — ■t t _ J 




49 
extension analyses of bovine mitochondrial nucleic acid, we 
also detected a series of DNA 5' ends near the putative 
porcine light strand transcription initiation site. 
Although the identified porcine DNA ends lie upstream of the 
RNA start site, they are also the probable result of nicks 
in the mitochondrial DNA. 

Attempts to determine the exact nucleotide of the major 
L-strand RNA 5' end were hindered by sequence differences 
between our cloned DNA and the mitochondrial RNAs of some of 
the individual pigs tested. Most notably, an eleven base 
pair insertion (CTTATAAAACA) which exists as a direct tandem 
repeat is located seven base pairs downstream from the major 
5' end identified above. Although this insertion appears in 
the DNA clone, it is not in the in vivo RNAs analyzed. To 
overcome this problem we devised a "run-off" dideoxy 
sequencing technique to directly sequence the mitochondrial 
RNA. A 32 P end labeled oligonucleotide was annealed to 
total mitochondrial RNA and primer extended using dideoxy 
sequencing mixes. This procedure resulted in a readable 
cDNA sequence copy of the RNA which terminated at the 5 ' end 
of the transcript, base 1139 on the porcine sequence (Figure 
3-4, Panel B) . Thus, the porcine in vivo L-strand RNA start 



site was mapped to this location. 





















50 



Sequence Alignment of the Bovine and Porcine 5 1 D-loop 
Regions 

A sequence alignment of the bovine and porcine 5' D- 

loop regions using their respective CSB-1 elements, L-strand 

RNA start sites, and phenylalanine tRNA genes as points of 

reference is shown in Figure 3-5. Only sporadic sequence 

similarity is seen in this region between these two 

Artiodactyl species, with the pig D-loop region containing 

an extra 354 bp not found in the cow. Immediately upstream 

(3 1 ) from the porcine CSB-1 (position 784) begins a stretch 

of 357 bp that is not found in the bovine D-loop. Only 

about half (160 bp) of this extra sequence, however, can be 

attributed to a domain containing tandemly repeated copies 

of the 10 bp sequence: CGCTGCTACA. The remaining half 

contains CSBs-2 and -3, two elements which had been 

previously suggested to be absent from this region of the 

bovine D-loop (Anderson et al. 1982). Interestingly, the 

only bovine sequence between CSB-1 and the light strand RNA 

start site that resembles a porcine sequence is a 37 bp 

domain comprised of two runs of C residues interspersed with 

an A/T rich stretch (positions 216-238) . The relative 

position of this bovine sequence block between CSB-1 and the 

light strand RNA start site and its partial similarity to 

both CSB-2 and CSB-3 (Figure 3-7) , suggest that this region 

may be functionally analogous to the consensus CSBs -2 and - 

3 seen in all other vertebrates. For these reasons and 

others discussed below, we have labeled this element "CSB- 



Figure 3-5. Sequence alignment of the 5' ends of the bovine 
and porcine D-loop regions . The light strand D-loop DNA 
sequences of the pig (top) and cow (bottom) were aligned 
relative to their respective CSB-1 elements, L-strand 
transcriptional start sties and phenylalanine tRNA genes 
(tRNAphe) . Consensus CSBs-1, -2, and -3 and a proposed 
analogue of the CSB-2 and -3 elements found in the bovine 
sequence, designated CSB-2+3 (see text) are indicated by 
brackets. Origins of H-strand DNA replication (OH) and L- 
strand transcriptional start sites (LSP) are indicated by 
arrows showing the direction of synthesis. The 10 bp repeat 
region in the porcine sequence is underlined, with spaces 
separating monomer units. Heavy lines indicate regions of 
similarity between the porcine sequence at positions 1158- 
1177 and a region of the bovine sequence previously 
suggested to be a CSB-2 element. Colons indicate sites of 
sequence homology while hyphenated regions show gaps in the 
two sequences . 



52 



4— I OH 

I 740 



680 700 720 

ATGGC-GTCAAAGGCCCTAACACAGTCAAATCAATTGTAGCTGGACTTCATGGAACTCATGATCCGGCACGACAA 

• •••• itiiiiiiitsi a * * * iti iftftftifttftti * * • ' * * ■ ■ ■ • • • ■ • 

ATGGCCGTCAAAGGCCCTGACCCGGAGC-ATCTATTGTAGCTGGACTTAACTGCA-TCTTGAGCAC-CAGCATAA 

100 4_J0H 140 xJoH 160 

I CSB-1 1 

780 800 

TCCAAACAAGGTGCTATT-CAGTCAATGGTTACGGGACATAA CGTGCGTACA CGTGCGTACA CGTGCGTACA 



4-1 



TGATAAGCATCCAC-ATTACAGTCAATGGTCACAGGACATAA 

OH I CSB-1 ' 

820 860 880 

CGTGCGTACA CGTGCGTACA CGTGCGTACA CGTGCGTACA CGTGCGTACA CGTGCGTACA CGTGCGTACA 



900 920 940 

CGTGCGTAC A CGTGC GTACA CG TGCG TACA CGTGCGTACA CGTGCGTACA CGTG CGTACA CGTCGTACA 



«™ i CSB-2 ■ 

960 980 1000 1020 

CGCGCATATAAGCAGGTAAA1TATTAGCTCA1TCAAACCCCCCTTACCCCCCATTAAACTTATGCTCTACACAC 



I CSB-3 1 

p.160 1080 1100 

CCTATAACGCCTTGCCAAACCCCAAAAACAAAGCAGAGTGTACAAATACAATAAGCCTAACTTACACTAAACAAC 



1120 1140 

ATTTAACAACACAAACCACCATATCTTATAAAACACTTA- 



ATTATArrATATATCCCCCCTTCATAAAAATTTCCCC 
I 220 I 

4— ILSP ' CSB 2 + 3 > 

! 116 ° 1180 1200 

CTTAAATACGTGCTACGAMGCAGGCACC^ 

CTTAAATA--T-CTACC ACCACrTTTAACAGACTTTTCCCTAGATACrrATTTAA 

^-JLSP 260 280 

1220 »«> 1260 1280 

AATTACAACACAATAACCTCCCAAAATATAAGCACCTATTTAAGCATACGCCCACAATCTGAATATAGCT 

'• ! : s : I » l : : ! : t i : : : s » i : : : : mm 8 

ATTTTT--CAC GCTTTCAATA CTCA-A1TTA-GCACTC-CAAACAAAGTCAATATAT— 

tRNAphe • 32 ° 340 

I 1300 1320 

TATAGTTAATGTAGCTTAAATTATCAAAGCAAGGCACTGAAAATGCCTAGATGG 

j^AOGCAGGCCC^CCC^----GTTGATGTAGciTA 

^^ I 380 400 

1340 tRNAphe 

GCCTCACAG— CCATAAACA 
: : : : ll ::::::::: 

GTCTCCCAACTCCATAAACA 
420 



53 
2+3". From the L-strand RNA start sites at positions 1139 
and 240, for the pig and cow respectively, to the 
phenylalanine tRNA genes of these species, the two sequences 
show only limited similarity. 

Identification of Putative Light Strand Transcriptional 
Regulatory Domains by Their Evolutionary Conservation 

An alignment of the mitochondrial D-loop DNA sequences 

of four Artiodactyls (cow, water buffalo, giraffe, and pig) 

relative to the 5 ' ends of the bovine and porcine in vivo L- 

strand transcripts is shown in Figure 3-6. Some striking 

interspecies sequence similarities and differences are noted 

in the figure. These features were used to divide the 

region into a series of putative regulatory domains (P, A, 

B, C, and D) . A comparison of the sequences encompassing 

the L-strand transcription start sites of all four species 

shows an eight base pair sequence 5 ' -CTTAAATA (domain P seen 

in Figure 3-6) which is completely conserved. This strict 

homology at all putative initiation sites suggests a 

possible function for this domain as a proximal core 

promoter sequence, and it has been therefore labeled domain 

P. Immediately 5' to domain P is another conserved region 

of 5 base pairs designated domain A. We used the GTG 

insertion seen in the pig sequence as well as other less 

obvious sequence differences seen in all species as a 

criterion for separating domains P and A. Just upstream 

from domain A, however, is a 20 base pair insertion seen 



54 
only in the pig. Although a seguence block very similar to 
this region is contained within the bovine mitochondrial 
genome (Figure 3-5), it is located at the extreme 5' end of 
the D-loop region and abuts the phenylalanine tRNA gene. 
Interestingly, in the cow this block lies about 5 bases 
downstream of an identified heavy strand transcriptional 
start site (King and Low 1987) . If this seguence has a 
function in transcription, it must operate at the LSP in the 
pig and at the HSP in cow, a seemingly unlikely scenario. 
As above, we used the porcine 2 base insertion seguence as 
a reference to delineate a boundary between conserved 
seguence elements. A remarkable feature of the next 
element, domain B, in the cow is a 12 base pair region 
(located at position -14 to -25) identical to the seguence 
seen at positions -11 to -22 relative to the light strand 
RNA start site in the human mitochondrial D-loop (Figure 3- 
6) . Further upstream from the putative core promoter other 
regions of seguence conservation, labeled C and D in the 
figure, were also seen. Subtle interspecies seguence 
variations were used for division of the two regions as well 
as for the upstream boundary for domain D. 

Immediately downstream from domain P, the DNA seguences 
of the cow, water buffalo, and giraffe diverge significantly 
from that of the pig. All but the pig seguence contain runs 
of C residues interspersed by stretches of A's and/or T's 
(in the light strand DNA seguence) in this region. Although 



0) o u 


> +J <U T3 >iTJ 


•h a) a) +j o) ja c 


P Xi >i Q) N (0 


(0 +J C-PWg-HT) 


-P-H • «H «-» O 0) • H 


X!<U <W 01 -H fc b 0) X (X 


tJiM(PO-H4JiflI/10.J5C 


•H W (0 rH d) +J«JO 


JT)CCI1I>'H'HT3CH0 


Q)Q)ox:Meoc >i<w w 


• c3-H+Jd)-Ha)tow 


<; en tr-p c oi oi a u - a) 


•H 0)-H >, C 10 -P 0) -— XS 


H rH 0) 01 01 O TJ t/10)Q-P 


o)(0 o oa)-H tj + 


c cu a <m oi >. -p -h u o 


njsx! o <u o p .c -p -p 


0< OP 0) UftU D'O'OC 


o x:cco(t3-HQ)c-H 


c+Jocu^'O-H.-iia 


• tj-h -h 3 a o o tjh 


U) 


C 01 P O 1 -H H 3 ^ 11 1 


Tl 


(Oioa)oa)x:p>iCcQCvD 


n 


CUP 0) 10 4-> U P o o 


i 


O(0«3M -H<ocr«-rHQ) 




rH O O -H <M > 1 <TJ-H<0*-I 


- 


(0 -HTiO C TJ <H 3 


in 


<u T3 CO OO «>itJ> 




<*-l • C <D 0) O C -H (X-H-H 


< 


3 C7>-H .£ C -H P<^PCl, 


■z, 


Xi -H P O tP >-i M Z c 


& 


Q, £ -HQ)Q}<;QlOa)C 




>4 OTJCT'MX: C3-H 


T3 


0) T3 k C 0) P U >i-H D 1 


C 


PCMrtJ^KJO-HJ-irtJCUC 


(0 


(0 rd f0 P (G g 0) 5 


^ 


5 WQ)L0T!a)POJ3O 


P 


S J) 0) C ill C£ CtJ 3£ 


oi 


-O-C-P-HCflPO) 0) 01 


1 


CUUE-" -H^H-H C gft 


hJ 


<u iopMCr>>ia)coioio 




4-1 0) 3+J-r(l/lHj(0(fl 


.H 


m£ -4J o d a a s 


> 


>_i P oi >-i o ill en u 


P 


•-H OlfdQ 0) .C O >i< O 


o 


CT<w XlP + nilHOPZP 


(0 


O 01 01 U + P OQU 


TS 


<fl cn 10 (0 0) 


O 


tP 01 TJ < TJ C • P TJ 01 > 


•H 


•hq) 2 c cq -h m c o -h 


-P 


Q,P >.2 <0 W 0) -H .£ -P 


M 


•HU UC^HOIPBD- 


< 


alio t» ~ a) a) a> ^ -h 




,£ TJCCQ OJC^rt: >-i 


<«-t 


p+jcuto • w id a • o 


O 


mp^ -to o« a) tj oi oi 




4-1 <0 C P < 0) r> hill 


P 


O P CD 01 -HI > P 3 


C 


0) 01 1 -O .QJ^-HrH 


a) 


ui aiijcucu'd'^uaioim 


e 


<d .< m acr-coi 


c 


u^aaioioifci m c m m 


o 


caaicc ndos^: 
ai M-H-H^fM trosp 


•H 


iH 


p TJ OrddlCUCU N 


< 


trca)Mgoo)^uia)G<w 


Q)n3MOO<WCQ3 XIQJO 


oi u (0 aT3 co tn a) P 


+J M U -H x: CM 


(D< 01 IIITJ'OH feHCOO) 


1 15 1 P G <D rtJ 01 -H-Hp 


nQJC«4J 3TJ nJ-P-H 


cu id c 10 c •pun 


ai-ocuscuc-HCfOcc-H 


hC^CCC 0) OOMH 


3 «J -P C7>-H -H G 01 -P -H O -P 


o> ^ -h > oi a> c X tr> oirH 


•H4->OrHO(l)Cl)OQ)<l)OQ)r0 


fcwpnjjaQWU-PM-P^W 







56 







P t" [• K 
H l< [• [< 
l< H H H 

I • 1 1 I. I . 
ti H O U 
<*< U 



m 



n 



< H H K 

1333 

. H H |. 



13 



H H * 

! • < I ■ 
OO t) 
u u o 

!'!•[. 
H O O 
H O O 
OUO 
O O O 
OO <J 

* o u 
OO o 
OH H 

o 4 H 

< K O 



m 



57 

the pig sequence similarly contains no G nucleotides in this 
area, the comparative organization consists of fewer 
homopolymer runs. This lack of apparent sequence 
conservation downstream from the putative core promoter 
sequence suggests that the sequence of bases downstream from 
position +3 (relative to the transcriptional start site) is 
not critical to the regulation of transcription. An 
interesting feature of the downstream C rich region shown to 
be conserved among cow, water buffalo, and giraffe is the 
similarity that these sequences have to conserved elements 
CSB-2 and CSB-3 seen in all other vertebrate species (Figure 
3-7) . While this region resembles neither CSB-2 or -3 very 
closely it contains features of both which suggest a 
functional similarity. What perhaps makes this region even 
more interesting is the absence of consensus CSBs-2 and -3 
from only these three species. Therefore, we have also 
labeled this region CSB 2+3 in the water buffalo and 
giraffe. 

Discussion 
We have mapped the major L-strand RNA 5' ends of both 
the cow and pig, two of the more distantly related species 
within the Artiodactyl order (approximately 52 million years 
divergent) , to similar positions in their respective D-loop 
regions. These sites in both species are located upstream 
from CSB-1, and CSBs-2 and -3 in the pig, and from CSB-1 and 



58 
a CSB 2/3 like element in the cow. Both are about 100 
nucleotides downstream from the phenylalanine tRNA gene. 
The positions of these RNA 5 ■ ends are analogous to the 
major light strand start sites seen in all other vertebrates 
analyzed. Furthermore, in the cow, we have also mapped the 
5 ■ end of a D-loop RNA species of opposite polarity to the 
same position as the L-strand RNA 5 1 end. This is only the 
second in vivo RNA which has been shown to be derived from a 
bidirectional LSP element. 

An alignment of the porcine and bovine D-loop 
seguences relative to their L-strand RNA 5 ' ends shows 
surprisingly little seguence conservation between these two 
related species. While both maintain the 8 bp seguence 
CTTAAATA (domain P) encompassing their major L-strand RNA 5' 
ends, only 10 bp upstream from this site the pig seguence 
contains a 20 bp insertion relative to the cow. Seguences 
further upstream show sporadic but limited homology. An 
alignment of the D-loop seguences of these two species with 
other Artiodactyls (water buffalo and giraffe) intermediate 
in evolutionary divergence between pig and cow, shows that 
domain P is strictly conserved in all four species. Several 
blocks of more limited seguence homology are also seen 
upstream from this region. 

Are Mapped RNA 5 ' Ends Actual L-strand RNA Start Sites? 

Several lines of evidence indicate that we have mapped 









59 
the L-strand start sites of four Artiodactyl species. 
First, the L-strand RNA 5 1 ends of both pig and cow map to 
similar regions within their respective D-loops, about 100 
bp downstream from the phenylalanine tRNA gene, and upstream 
from the heavy strand replication origin (OH), CSB-1, CSB-2, 
and CSB-3 in the pig, and CSB-1 and a CSB-2/-3 like region 
in the cow. The pig and cow sites are also consistent in 
location with the L-strand start sites of all other 
vertebrates analyzed. The positioning of a transcription 
initiation site upstream from the OH and CSBs -1, -2, and -3 
is believed to be critical for the priming of mitochondrial 
DNA replication (Gillum and Clayton 1979; Chang and Clayton 
1985; Chang et al. 1985) in that the CSBs are believed to be 
necessary recognition domains for the transition from RNA to 
DNA synthesis (Chang and Clayton 1987) . Additionally, we 
have identified bovine H-strand RNAs whose 5' ends map to 
identical positions as the major L-strand RNA 5' ends. 
Bidirectional initiation of transcription, first seen in 
vitro and later documented in vivo in the human (Chang et 
al. 1986; Bogenhagen and Romenelli 1988) , is believed to be 
characteristic of D-loop promoter sequences in vertebrate 
mitochondria. Thus, the existence of bovine RNAs of 
opposite polarity with common 5 ■ end positions leads us to 
conclude that we have identified a bidirectional D-loop 
transcriptional start site in the cow. The similar map 
position and sequences surrounding the porcine major L- 



60 
strand 5 ■ end identified here suggest that we can extend 
these conclusions to the pig as well, and by sequence 
conservation of domain P and regions upstream, by inference, 
to water buffalo and giraffe. 

Sequence Similarities at Artiodactyl L-Strand Initiation 
Sites Suggest Conserved Transcriptional Regulatory Functions 

The evolutionary conservation of sequences at or near 

sites of transcription initiation suggests their role as 

regulatory domains for RNA synthesis. When the D-loop 

sequences of other related Artiodactyl species (water 

buffalo and giraffe) were aligned relative to the L-strand 

start sites of the pig and cow, we were able to identify 

several domains of sequence conservation in all four 

species. Perhaps the most consistent theme found throughout 

the regions near and upstream from the Artiodactyl RNA start 

sites is the conservation of nucleotide composition. The L- 

strand DNA sequences are A-T rich (about 74%) and nearly 

devoid of G residues (about 6%) . The analogous region in 

the mouse is also A-T rich (77%) , and both human and mouse 

LSP regions are also deficient in G residues (13% and 11% 

respectively) . Nucleotide composition of this nature may 

reflect a relevant aspect of the makeup of regulatory 

regions in the mitochondrial LSP region, that is the ability 

to easily denature due to high A+T content as a prelude to 

transcription. At the nucleotide sequence level, the most 

notable block of sequence homology between the four 



61 
Artiodactyl species is the strictly conserved 8 bp sequence: 
CTTAAATA which contains the L-strand start sites of all four 
species. This domain is the largest uninterrupted stretch 
of sequence to be conserved in these species in this genome 
region. 

In vitro transcription studies in human and mouse have 
suggested that mitochondrial promoters consist of at least 
two functional parts: a sequence required for accurate 
initiation and an upstream domain involved in regulating 
transcription efficiency (Chang and Clayton 1984; Chang and 
Clayton 1986b; Chang and Clayton 1986c) . The extensive 
sequence heterogeneity between the species, however, has 
hindered elucidation of the exact size, location, and 
relevant sequences of these elements. The strict sequence 
conservation seen only at the Artiodactyl L-strand RNA start 
sites strongly suggests that we have identified a core light 
strand promoter sequence or at least a principal component 
of the initiation domain. Interestingly, the sequence 
homology at this domain diminishes 3 bp downstream from the 
initiation nucleotide, which suggests the upstream boundary 
of the Artiodactyl light strand promoter region. 

Upstream from the core element we note other blocks of 
sequences (labeled A, B, C, and D in Figure 3-6) that are 
somewhat less conserved in all four species. Contained 
within domain B in the cow is the 12 bp sequence: 
CACTTTTAACAG, which is also found in the human LSP in an 



62 

analogous position. In the human, this sequence has been 
found to be part of a larger (about 4 bp) sequence bound by 
the human mitochondrial transcription factor (mitTF 1) 
(Fisher et al. 1987) . Its presence in the cow and the 
existence of similar sequences in domain B of the other 
Artiodactyls, strongly suggests that this region is 
important to transcription in these species. 

The sequence alignment of the four Artiodactyl species 
allows us to identify the occurrence of an apparent 20 bp 
insertion event in the pig sequence 10 bp upstream from the 
L-strand start site. That this region is indeed an 
insertion is suggested first by the conservation seen 
between all four species flanking the 20 bp region, and 
second by the identification of a sequence 85% identical, 
positioned in the cow immediately adjacent to the bovine 
phenylalanine tRNA gene. The position and maintenance of 
the 20 bp sequence just 5' from the L-strand RNA start site 
raises several questions about mitochondrial transcription 
and perhaps allows unique insight into regulation of this 
process. As mentioned above, in vitro transcription 
experiments have suggested that two domains are necessary 
for accurate and efficient transcription initiation; the 
porcine insertion may serve to separate and define the 
boundaries of these regions, suggesting that the position 
of, and distance between these two elements with respect to 
one another may not be critical for in vivo function. An 



63 

alternative possibility may be that the 20 bp region is in 
fact involved in porcine transcription. This would imply 
that the proteins involved in recognizing regulatory domains 
in this region can either identify a widely diverse range of 
seguences or that the proteins and cognate DNA seguences in 
the transcriptional apparatus have rapidly evolved and are 
species specific. A third alternative is that mitochondria 
have several options for activation domains and only one 
need be present to stimulate transcription in vivo . 
According to this scenario, the porcine 20 bp insertion may 
serve that role for the LSP in pig, and its analogue in cow 
serves that role for the HSP. If so it would constitute a 
unigue form of regulatory domain shuffling. 

Do Artiodactyl Species Contain CSBs-2 and -3 ? 

All vertebrates presently studied outside the 
Artiodactyl order contain three conserved domains, CSB-1, 
-2, and -3, positioned between the L-strand promoter and the 
D-loop DNA 5* end(s) (Anderson et al. 1981; Bibb et al. 
1981; Anderson et al 1982) . Unlike all other species within 
the Artiodactyl order thus far examined, the pig retains 
this arrangement. Previously, the cow was believed to have 
lost CSB-3 entirely, and a homopolymer C domain positioned 
adjacent to the bovine phenylalanine tRNA gene (nucleotides 
348 to 363) had been labeled as a potential CSB-2 (Anderson 
et al. 1982). As shown in Figure 3-5, this bovine poly-C 









64 
element appears to have an analogue, quite distinct from 
CSB-2, in the pig D-loop region suggesting that if a bovine 
equivalent of CSB-2 exists, it must lie elsewhere. Several 
lines of evidence lead us to suggest that more plausible 
Artiodactyl equivalents for CSB-2 and CSB-3 exist as 
stretches of sequence containing two domains of poly C 
residues located immediately downstream from the L-strand 
start sites in these species. Similar to CSB-2 
and CSB-3 in non-Artiodactyl species, these domains are also 
upstream from CSB-1. As seen in Figure 3-7 the runs of C's 
share sequence homologies to parts of both CSB-2 and CSB-3 
and are spaced such that they align precisely with poly-C 
residues within the proposed recognition sequences of the 
RNA component of mouse RNAse MRP (Bennett and Clayton 1990) , 
a site-specific ribo-endonuclease which uses CSB-2 and CSB-3 
as recognition sites (see above) (Bennett and Clayton 1990) . 
In Figures 3-5 and 3-6 we have labeled this region CSB-2+3 
because it appears to be a hybrid of the two elements, 
possibly representing minimal sequence domains required for 
its functional recognition. Thus, it may be that the 
critical cis-element satisfied by CSB-2 and -3 in most 
vertebrates are only L-strand poly-C runs and their position 
between the L-strand promoter and H-strand origin. This idea 
is consistent with the analogous placement of an L-strand 
poly-C run in sea urchin mitDNA which lacks any consensus 
vertebrate CSB-2 (Jacob 1984) . We therefore suggest that 












Figure 3-7. Alignment of the Artiodactvl CSB-2+3 region 
with consensus CSB-2 and -3 sequences and the recognition 
region of the murine RNAse MRP . The proposed CSB-2+3 
elements of the cow, water buffalo, and giraffe are shown as 
designated. Non-Artiodactyl consensus CSB-2 and CSB-3 
regions are aligned above. The recognition sequence of the 
RNA portion of the murine RNAse MRP is shown underneath 
(Bennett and Clayton 1990) with sites of sequence similarity 
to the bovine CSB-2+3 element indicated by colons. Sites of 
predicted interaction of the RNAse MRP with the murine 
mitRNA are underlined. The primary site of RNA cleavage 
within murine L-strand transcripts is shown by the arrow. 



66 



CSB-2 CSB-3 ' 



CCCCCCACCCCC TGCCAAACCCCAAAAACA 
GIRAFFE ACCCACCCTTTCCTTTCCCCCCCCCCC 
WATER BUFFALO CTTCCCCCCCTCCTATCAAA-CCCCCC 
COW TATCCCCCCCTCCATAAAAATTTCCCC 



RNAse MRP TAGACTTCCCCCGCAAGTCACTGTTAGCCCGCCAAGAAG 



t 






67 

CSBs-2 and -3 actually exist in the cow and its close 
relatives, but in a disguised form which retains only the 
functional features of the well studied consensus CSB-2 and 
-3, and that the minimal function of these CSBs is supplied 
simply by homopolymer C runs positioned appropriately 
relative to the LSP, OH and CSB-1. 









CHAPTER 4 

ISOLATION AND CHARACTERIZATION OF BOVINE AND PORCINE 

MITOCHONDRIAL RNA POLYMERASE ACTIVITIES 

Introduction 
The D-loop sequence heterogeneity that exists between 
unrelated species has prohibited the determination of DNA 
sequences involved in the regulation of mitochondrial 
transcription from a comparison of regions flanking initiation 
sites. Parallel investigations in a variety of species have 
attempted to define the interaction of the mitochondrial 
transcriptional machinery with its homologous DNA template. 
Due to the lack of an efficient method for introducing foreign 
DNA into vertebrate mitochondria, these studies have involved 
the isolation and analysis of protein extracts from purified 
mitochondria capable of transcription in vitro (Walberg and 
Clayton 1983; Chang and Clayton 1986; Bogenhagen and Yoza 
1986) . Mitochondrial in vitro transcription systems from the 
mouse (Chang and Clayton 1986) and human (Walberg and Clayton 
1983) have been isolated from tumor cell lines (L-cells and 
HeLa cells, respectively) grown in culture, while in Xenopus 
(Bogenhagen and Yoza 1986) extracts were made from oocytes 
stimulated into rapid division by gonadotropin. All three in 
vitro systems have been used to describe their respective 



69 
promoter elements by the ability of the system to accurately 
transcribe cloned mutant DNA templates. These investigations 
have demonstrated that, at least in vitro , fundamentally 
different promoter structures exist between mammals and 
amphibians. Mammalian mitochondrial promoters appear to 
consist of two sequence domains, one required for accurate 
initiation at the start sites and an upstream region involved 
in regulating transcription efficiency (Chang and Clayton 
1984; Chang and Clayton 1986a; Chang and Clayton 1986b) . In 
contrast, amphibians require only a single short 18 bp 
sequence flanking the initiation site for efficient 
transcription initiation in vitro (Bogenhagen and Romenelli 
1988) . For both mammalian and amphibian mitochondria, at 
least one transcription factor, separable from RNA polymerase 
activity, has been shown to be required for efficient 
initiation at designated start sites (Fisher and Clayton 1985; 
Bogenhagen and Insdorf 1988) . In mammalian systems this factor 
has been shown to be a DNA binding protein (Fisher et al. 
1987) . 

From an alignment of DNAs found upstream of the L-strand 
initiation sites of four related Artiodactyl species (cow, 
water buffalo, giraffe, and pig) , we have been able to 
identify domains which are evolutionally conserved in 
sequence, location and function in vivo (Chapter 3). In order 
to test the in vitro function of these sequences in more 
detail with regard to regulating mitochondrial transcription, 






70 
we have isolated extracts containing mitochondrial RNA 
polymerase activities from adult brain tissue of two of the 
most divergent Artiodactyl species, cow and pig. These 
extracts were then used for in vitro transcription assays and 
later in protein :DNA interaction studies (Chapter 5) to better 
define mitochondrial promoter elements. By developing in 
vitro transcription systems from normal animal tissue we have 
tried to avoid any effects which may be unigue to tumor cells 
or cells stimulated in an artificial manner. 

Results of transcription assays using the RNA polymerase 
activities from both pig and cow on their respective 
homologous templates show that the porcine RNA polymerase 
preparation is able to recognize and initiate transcription 
from the L-strand RNA start site identified from in vivo 
mitochondrial RNA. The bovine RNA polymerase, however, while 
able to initiate RNA synthesis at the major L-strand RNA start 
site identified in vivo , prefers to initiate in vitro at a 
position about 19 bases further upstream. This -19 location 
coincides with a site previously shown to be a minor in vivo 
L-strand RNA 5' end (Chapter 3). Even though the putative 
bovine and porcine LSP seguences are somewhat dissimilar, in 
heterologous transcription assays the bovine and porcine RNA 
polymerase activities are functionally interchangeable, each 
transcribing the LSP template of the other species in a manner 
indistinguishable from homologous in vitro transcription. 
These studies suggest that in spite of apparent seguence 









71 
differences found at putative transcriptional regulatory 
regions, the proteins involved in mitochondrial transcription 
in Artiodactyl species must be very similar. Furthermore, at 
least in vitro , these activities are not strictly limited to 
recognition of their cognate promoter seguences. These 
findings also indicate that further characterization of the 
components involved in transcription and their interaction 
with promoter seguences will be necessary to fully understand 
vertebrate mitochondrial promoter elements. 

Results 
Our initial attempts to isolate mitochondrial RNA 
polymerase from bovine brain tissue employed procedures 
similar to those described by Walberg and Clayton (1983) for 
preparation of mitochondrial RNA polymerase from human KB 
tissue culture cells. One to two grams of purified 
mitochondria were lysed in a 4 ml solution of column buffer 
(see materials and methods) containing 1% Triton X-100 and 
0.35 M KC1. The lysate was cleared by centrifugation and 
mixed with an egual volume of solution containing 2 mis of 
pre-prepared heparin-Sepharose in column buffer. The mixture 
was poured into a 1 cm X 8 cm column and washed with column 
buffer plus 0.175 M KC1. Proteins bound by the column were 
eluted with a 20 ml linear gradient between 0.175 and 0.6 M 
KC1 and 1 ml fractions collected. The fractions were assayed 
for non-specific RNA polymerase activity and active fractions 






72 
dialyzed into column buffer including 50% glycerol. Similar 
to the human system, this type of procedure routinely gave a 
broad peak of nonspecific RNA polymerase activity that eluted 
from the column between 0.3 5 M and 0.5 M KC1. The pooled 
activities or peak activity fractions were then incubated in 
run-off transcription assays using a cloned DNA template 
(pR16) containing bases 16202-457 of the bovine mitochondrial 
genome linearized at the Bam HI site at position 16202 
(diagrammed in Figure 3-1, Panel B) . A run-off transcript of 
376 bases would be generated if in vitro initiation occurred 
at the light strand transcriptional start site at nucleotide 
241. As seen in Figure 4-1, panel A, two prominent 
radioactive species were generated in the reaction migrating 
near positions 360 and 492. In early experiments we believed 
these bands to be specific transcripts initiating at the light 
strand promoter and an upstream bidirectional heavy strand 
promoter, near positions 376 and 476 respectively relative to 
the Bam HI site. However, upon testing the run-off reaction 
products for sensitivity to DNA and RNA specific nucleases, 
we found that both labeled species were DNAse sensitive and 
RNAse resistant (Figure 4-1, Panel A, lanes 4 and 3 
respectively) , thus, neither species was an in vitro 
synthesized transcript. The results from these assays 
indicated that the bovine mitochondrial protein fractions 
could digest the DNA template at specific sites and use 32 P UTP 
to label the fragments. To confirm this suggested presence 



Figure 4-1. Nuclease activity in heparin-Sepharose column 
fractions . Panel A. DNA nuclease activity. The R16 
plasmid (containing a cloned fragment of the bovine D-loop 
corresponding to nucleotides 16202-457, diagrammed in Figure 
3-1) was linearized at the Bam HI site at position 16202 and 
incubated in run-off transcription reactions (see materials 
and methods) with fractions from heparin-Sepharose columns 
containing bovine (lanes 1, 3, and 4) and porcine (lane 2) 
mitochondrial RNA polymerase activity. The reaction 
products were then electrophoresed on a sequencing gel. In 
lane 3 reaction products were digested with RNAse Tl prior 
to electrophoresis. In lane 4 reaction products were 
digested with DNAse 1 prior to electrophoresis. 
Radiolabeled fragments of a Hpa II digest of pBR322 were 
electrophoresed in lane M as DNA size standards. Panel B. 
RNA nuclease activity. A fraction containing peak bovine 
mitochondrial RNA polymerase activity eluted from a heparin- 
sepharose column (lanes 2, 4, 6, and 8,) was incubated with 
radiolabeled transcripts from a T3 run-off reaction (using 
the same linearized template as in panel A) . Incubation 
times were 5 min (lanes 1 and 2) , 10 min (lanes 3 and 4) , 15 
min (lanes 5 and 6), 20 min (lanes 7 and 8). Lanes 2, 4, 6, 
and 8 had no added fraction. 



B. 



M12 3 4 12345678M 



74 





5 2 7- 



404- 



309- 



9 














i 




75 
of a contaminating DNA nuclease in these polymerase 
preparations, we incubated a set of heparin-Sepharose column 
fractions with the 592 base pair Bam Hl-Hinc II, mitDNA 
fragment (again corresponding to bases 16202-457) which had 
been end-labeled at the Bam HI site. When the reaction 
products were sized on sequencing gels, a peak of DNA nuclease 
activity was found that cofractionated with RNA polymerase 
activity. This nuclease digested the DNA probe in the same 
manner as seen in the run-off reactions (data not shown) . The 
same fractions were also tested for contaminating RNA 
nuclease. As shown in Figure 4-1, Panel B, a level of RNAse 
activity was found in the peak RNA polymerase fraction 
sufficient to completely digest a synthetic 620 nt transcript 
in 20 min. From these experiments it was apparent that in 
vitro transcription experiments would be impossible without 
further fractionation of mitochondrial proteins. 

We employed several different strategies of additional 
protein chromatography in an effort to separate nuclease 
activities from the mitochondrial transcription machinery. 
Similar to procedures used by Chang and Clayton (1986) for 
isolating mitochondrial RNA polymerase activity from murine 
tissue culture cells, we tried protein fractionation with 
phosphocellulose (Whatman Pll) . Cleared lysates, pools of 
nonspecific RNA polymerase activity from heparin-Sepharose 
columns and lysates passed over DEAE-Sephacel at 0.5 M KC1, 
were bound and fractionated over variously sized columns of 



76 
phosphocellulose. Unlike the mouse mitochondrial RNA 
polymerase, the non-specific RNA polymerase activity from the 
cow bound very weakly to the phosphocellulose column. Worse, 
the bovine activity still cofractionated with the DNA nuclease 
activity, both eluting from the column at about 100 mM KC1. 
From studies involving Xenopus mitochondrial RNA polymerase, 
Bogenhagen (1988; and personal communications) successfully 
removed nucleases using hydrophobic chromatography (Phenyl- 
Superose, Pharmacia) . This type of column was useful for 
removing both DNAse and RNAse activities from bovine 
mitochondrial protein preparations. However, the recovery of 
RNA polymerase was very poor and the resultant activity was 
quite unstable. We also tried FPLC fractionation using both 
Mono-Q and Mono-S columns (Pharmacia) and found that either 
DNA nuclease activity cofractionated with RNA polymerase 
activity (Mono-Q) , or the polymerase activity was lost during 
fractionation (Mono-S) . In a modified preparation of human 
mitochondrial RNA polymerase, Fisher and Clayton (1985) 
reported success in isolating an activity capable of specific 
homologous in vitro transcription by fractionation of 
mitochondrial proteins over DEAE-Sephacel. In an attempt to 
mirror this success we fractionated cleared bovine lysates 
over 25 ml columns of DEAE-Sephacel using both linear 
gradients (0.05-0.5 M KC1) and batch elution techniques (i.e. 
proteins which remained bound to the column at 50 mM KC1 were 
eluted by washing at 0.5 M KC1) . While this approach proved 



77 
to remove about 90% of the contaminating DNAse activity, 
recovery of RNA polymerase activity was again very poor. 

The results of the above fractionation experiments 
suggested that the bovine RNA polymerase would bind well to 
heparin-Sepharose but was being largely excluded from the 
DEAE-Sephacel column, possibly due to the presence of a large 
population of other types of proteins competing for column 
matrix binding sites. In order to increase the yield of RNA 
polymerase as well as prevent overloading each of the columns, 
we tried a strategy whereby much larger columns of heparin- 
Sepharose and DEAE-Sephacel were used. As a first 
purification step to enrich the lysate for RNA polymerase 
activity, we bound cleared lysate (in column buffer with 0.2 
M KC1) to a 60 ml heparin-Sepharose column. After washing the 
column with low salt (0.2 M KCl) , bound proteins were batch 
eluted with a step gradient at 0.6 M KCl, and fractions 
collected. Active fractions were then dialyzed in column 
buffer or diluted with column buffer to reduce the KCl 
concentration to about 0.05 M and loaded onto a 50 ml column 
of DEAE-Sephacel. The column was washed extensively at 0.05 
M KCl, and then to insure that all proteins involved in 
transcription eluted together, bound proteins were batch 
eluted at 0.5 M KCl. Protein fractions were collected and 
dialyzed individually into column buffer with 50% glycerol and 
stored at -2 0°C. 









78 
After dialysis, the fractions were assayed for 
nonspecific RNA polymerase activity. A single peak of 
activity was seen in fractions 6 through 8 with a maximum of 
about 25,000 cpm incorporation of radiolabeled UTP in fraction 
7 (Figure 4-2, Panel A). To assay for the elution of 
potential transcription factors, we also tested a subset of 
these fractions for their ability to retard the migration of 
a labeled 592 bp DNA fragment containing the 5' end of the 
bovine D-loop (nucleotides 16202-457 on the bovine seguence) 
in a series of band shift experiments (Figure 4-2, Panel B) . 
As expected for batch elution, we found a single peak of DNA 
binding activity in fractions 5 through 8, with maximum 
activity again in fraction 7. These preliminary assays 
indicated that the majority of RNA polymerase and DNA binding 
activities were concentrated in fraction 7. When the same 
fractions were tested for the ability to initiate specific 
transcription from a cloned DNA template (pR16, containing 
bases 16202-457 from the bovine sequence linearized at the Bam 
HI site, as above) in run-off transcription assays, no 
specific transcripts could be detected (data not shown) . 

In an effort to minimize the effects of any contaminating 
RNA nucleases in the protein fractions, we tried using a 
template which would allow the detection of much smaller run- 
off transcription products. A synthetic Artiodactyl LSP, 
directionally cloned into the Bam HI- Sail sites of the 
Bluescript vector (pALSP, diagrammed in Figures 3-6 and 6-1) , 



Figure 4-2. Elution of non-specific mitochondrial RNA 
polymerase and DNA-binding activities from the DEAE- 
Sephacel . Panel A. Non-specific RNA polymerase activity. 
Bovine mitochondrial proteins bound to a DEAE-Sephacel 
column (see text) were batch eluted and collected in 
fractions. Fractions 3-13 were assayed for non-specific RNA 
polymerase activity by their ability to incorporate 
radiolabeled nucleoside triphosphates into acid precipitable 
counts using denatured salmon sperm DNA. Panel B. Non- 
specific DNA binding activity. Fractions 5-12, from above, 
were assayed for their ability to retard the migration of a 
592 bp radiolabeled DNA fragment in DNA binding gels. The 
arrow indicates the position of migration for the unbound 
DNA probe. 












80 






30000 n 



< 
ac 

o 

§| 20000 

o 

u 



E 
a, 
u 



10000- 




o I ■ ■ I 



~r 



-i — i — | — i — i — | 



4 6 8 10 12 

FRACTION NUMBER 



14 



B. 



5 6 7 8 9 10 1112 






81 

was linearized at the downstream SST 1 site contained within 
the polylinker of the vector and incubated in run-off 
transcription reactions with the various column fractions. 
From this template, initiation of RNA synthesis from the 
consensus Artiodactyl start site would result in labeled RNA 
species of about 67 nucleotides in length. As shown in Figure 
4-3, a peak of activity was found in fractions 6 and 7 
(primarily in 7) , which was able to generate a single 
detectable labeled band. This band, however, was larger than 
expected, migrating at about 77 nucleotides relative to the 
DNA size standards. Fractionation of porcine mitochondrial 
proteins in parallel, using the same chromatographic 
procedures also allowed us to identify fractions (numbers 6 
and 7 in Figure 4-4) capable of producing two predominantly 
labeled species of about 113 and about 85 nucleotides 
(relative to DNA size standards) when incubated with a cloned 
linear template containing porcine D-loop seguences flanking 
L-strand initiation site (pBB21, containing nucleotides 1015 
to 1430 of the porcine seguence cloned into the Bluescribe 
vector (S. Mackay, Ph. D. thesis, University of Florida), 
linearized at the Pst 1 site adjacent to nucleotide 1015, 
(diagrammed in Figure 4-4, Panel B) ) . Relative to the DNA 
size marker, both these bands were smaller than the expected 
RNA size of 135 nt. 

Having generated discreet labeled RNA products from the 
run-off reactions, we attempted to optimize reaction 



Figure 4-3. Specific transcription activity in bovine 
mitochondrial protein fractions eluted from a DEAE-Sephacel 
column . Fractions 3-13, from above, were incubated in run- 
off transcription reactions with pALSP (linearized at the 
Sst 1 site) and the products electrophoresed on a sequencing 
gel. Numbers in the margin indicate approximate size from 
DNA size standards. 



83 



M3 4 5 6 78 910111213 



9 0. 

7 6- 

6 7_ 



3 4. 



^ 






Figure 4-4. Specific transcription activity in porcine 
mitochondrial protein fractions eluted from a DEAE-Sephacel 
column . Panel A. Run-off transcriptional activity. 
Fractions 4-12 (batch eluted from a DEAE-Sepahcel column 
similar to bovine extracts above) were incubated with the 
cloned DNA, pBB21 (diagrammed in Panel B) linearized at the 
Pst 1 site, in run-off transcription reactions. The 
products were electrophoresed on a seguencing gel. Numbers 
in the margin indicate approximate DNA sizes in nucleotides. 
Panel B. Diagram of pBB21. Nucleotides 1015-1430 of the 
porcine mitDNA seguence were cloned into the Bam HI and Pst 
1 sites of the Bluescribe vector (S. MacKay, dissertation, 
University of Florida) . This region contains the porcine 
CSB-3 element, the phenylalanine tRNA gene, and a portion of 
the 12S rRNA gene (all are indicated as labeled) . The major 
site of in vivo L-strand transcription initiation and 
direction of synthesis is indicated by the arrow labeled 
LSP. Approximate nucleotide numbers of the porcine mitDNA 
are indicated. 



85 






9 10 11 12 




B. 



CSB-3 



PST 1 



— 1 — 

1100 



LSP 



~l — 

1200 



- r 

1300 



tRNAPHE 12S rRNA 
■ I . ' 



1400 Bam HI 



86 

conditions, varying such reaction parameters as pH, 
concentrations of MgCl2, MnC12, KC1 and template DNA as well 
as time and temperature of incubation. A representative 
example is shown in Figure 4-7, Panel A where we tested the 
ability of the porcine RNA polymerase activity to transcribe 
its homologous cloned template at various concentrations of 
MgCl 2 and MnCl 2 . From these experiments we found that 
conditions similar to those used for the human and mouse 
systems also worked well for the bovine and porcine extracts. 
A series of experiments using the same templates (for the cow, 
the Artiodactyl LSP clone was linearized at the Xho 1 site 
contained within the Bluescript polylinker) showed that the 
reaction products from both species 1 extracts are resistant 
to digestion with DNAse 1, sensitive to RNAse Tl and dependent 
upon the addition of both template and mitochondrial protein 
fraction (Figure 4-5) . From these results we concluded that 
both bovine and porcine mitochondrial protein preparations 
were able to synthesize RNA in vitro . These experiments also 
revealed a faintly labeled RNA species which migrated at about 
105 nt. The presence of this band was not template dependent 
and was found only in reactions where either the bovine or 
porcine polymerase activity had been added. Since this 
labeled species had no apparent effect on our experiments, we 
have not addressed its existence further and tentatively 
conclude that it represents an RNA which copurified with 
transcriptional activity and became labeled during the run- 









Figure 4-5. Sensitivity of run-off transcription products 
to digestion with RNA and DNA specific nucleases . Fraction 
7, containing the peak bovine run-off reaction activity 
(Figure 4-3) was incubated with pALSP (previously 
linearized at the Xba 1 site in run-off transcription 
reactions (lanes 1-4) and the products electrophoresed on a 
sequencing gel. Prior to electrophoresis lanes 3 was 
treated with RNAse Tl, while lane 2 was incubated with DNAse 
1. In lanes 6-9, fraction 7, containing the porcine run-off 
transcription activity (Figure 4-4) was incubated with pBB21 
(previously linearized at the Pst 1 site) in run-off 
transcription reactions. Prior to electrophoresis lane 8 
was incubated with RNAse Tl and lane 7 with DNAse 1. No 
fractions were added to the reactions shown in 5 and 10. No 
DNA template was present in lanes 4 and 9. 



88 



M 12 3 456789 10 







M 



89 
off reaction. We note that mitochondrial RNPs have been 
isolated (Chang and Clayton 1987) which contain specific RNAs. 
After identifying fractions capable of generating run- 
off transcripts of a discreet size, we switched to primer 
extension analyses to precisely determine initiation sites of 
the in vitro RNA. Although we were unsuccessful at using the 
native bovine template in run-off transcription experiments, 
we reasoned that positioning the annealing site for an 
oligonucleotide primer close to the transcriptional start site 
(for extension by reverse transcriptase) might allow detection 
of transcripts whose synthesis had been initiated, but did not 
extend through the full length of the DNA template. Also, 
since the source of radioactivity would be the end-labeled 
oligonucleotide primer rather than the in vitro synthesized 
RNA, we could use a higher concentration of UTP in the in 
vitro transcription reaction to promote better elongation. 
For this type of assay, the protein extracts were first 
incubated in transcription reactions with an undigested 
cloned, circular DNA template and the four unlabeled 
nucleoside triphosphates. After the reaction, the samples 
were extracted with phenol, ethanol precipitated and then, to 
remove template DNA, resuspended in a reaction volume 
containing an excess of DNAse 1. After DNAse digestion the 
samples were phenol extracted, precipitated with ethanol and 
annealed to a radiolabeled oligonucleotide primer for 
subseguent extension by reverse transcriptase. Aliquots of 



90 
in vivo bovine mitochondrial RNA were also pretreated with 
DNAse 1 and used in primer extension assays for comparison to 
the RNA synthesized in vitro . As shown in Figure 4-6 we were 
able to detect in vitro transcripts which had been synthesized 
on the bovine D-loop template. However, the most prominent 
in vitro RNA, initiated at a position about 19 nucleotides 
upstream from the major in vivo L-strand RNA 5' end. This 
-19 site is coincident with the 5 ' ends of minor in vivo L- 
strand transcripts (identified previously in Figure 3-2 and 
here in Figure 4-6, lane 7) . Although we can detect in vitro 
transcripts which initiate from the same position as the major 
in vivo RNA, they are somewhat less prominent. Much further 
upstream we also note several labeled species representing 
RNAs which initiate from a position consistent with 
bidirectional initiation from a putative heavy strand promoter 
region. In vivo bovine mitochondrial RNA 5' ends were also 
detected in this region (highlighted by the boxed region 
adjacent to lane 7) . 

Aliquots of the transcriptionally active porcine 
mitochondrial protein fraction were also incubated with the 
bovine DNA template in a series of heterologous transcription 
reactions. Primer extension of the in vitro reaction products 
showed that the porcine fraction transcribed the bovine 
template in a manner virtually identical to that of the bovine 
fraction (Figure 4-6, lanes l, 3, and 5). As shown in Figure 
4-6, an increase in the amount of each extract added to the 



Figure 4-6. Specific transcription of the bovine LSP region 
by the bovine and porcine mitochondrial RNA polymerase 
activities . Panel A. Primer extension of L-strand RNAs 
synthesized in vitro from the bovine D-loop region. 
Fractions containing bovine (lanes 2, 4, and 6) and porcine 
(lanes 1, 3, and 5) mitochondrial RNA polymerase activities 
(2 ul in lanes 1 and 2, 4 ul in lanes 3 and 4, 8 ul in lanes 
5 and 6) were incubated with pR16 (undigested) in 
transcription reactions. Following incubation with DNAse to 
remove template DNA, the RNA products were annealed to a 
radiolabeled oligonucleotide primer (corresponding to 
nucleotides 136-157) and extended with reverse 
transcriptase. The products were then electrophoresed on a 
sequencing gel. Lane 7 represents primer extension of in 
vivo bovine mitochondrial RNA. Lanes A, G, C, and T 
represent dideoxy sequencing reactions of the template DNA 
using the same primer as above. The dark arrow represents 
the major in vivo site of L-strand transcription initiation. 
The light arrow indicates the major site (-19) of in vitro 
transcription initiation. The box highlights bidirectional 
transcription from the bidirectional HSP region. The bovine 
CSB-2+3 element (see Chapter 3) and the phenylalanine tRNA 
gene are designated by boxes with dashed lines. Panel B. 
Sites of in vivo and in vitro L-strand transcription 
initiation. Bovine light strand DNA sequences near the L- 
strand transcription initiation site are shown. Major and 
minor sites of transcription initiation are indicated by 
dark and light arrows, respectively. Sites of in vivo and 
in vitro transcription initiation are as indicated. 



92 



A. 



A G C T 1 2 3 4 



4-> «- 



00 

+ 

m 
in 
U 




B. 



IN VIVO «L I 

ATAAAAATTCCCCCTTAAATATCTACCACCACTTTTAACAGACTTTTCCCTAGAT 
ttt f f 



IN VITRO 



93 
transcription reaction resulted in a similar increase in 
primer extension products. Interestingly, mirroring the 
bovine RNA polymerase, the porcine activity appeared able to 
identify the major in vivo site, but also favored initiation 
in vitro at the -19 site. The porcine activity also initiated 
from the putative bidirectional heavy strand promoter region. 
When the porcine and bovine RNA polymerase activities were 
used to transcribe the synthetic Artiodactyl template, both 
initiated transcription from a region corresponding to the - 
19 site seen on the bovine template. This accounts for the 
larger than expected in vitro RNA species seen in the original 
assay of the bovine protein fractions seen in Figure 4-3. It 
is important to note that in general, we found that the small 
in vitro run-off RNAs made in these experiments migrated at 
only about 80-85% of their actual size relative to the DNA 
size standard used and conclude that primer extension is 
required to accurately size and map run-off transcripts. 

In experiments analogous to those above, primer extension 
analyses were used to accurately map the 5' ends of RNAs 
transcribed in vitro from the porcine template. We found that 
both porcine and bovine polymerase activities transcribed the 
porcine D-loop template in precisely the same manner 
identifying the same major and minor initiation sites, both 
in run-off transcription (Figure 4-7) and primer extension 
(Figure 4-8) reactions. As shown in Figure 4-8, both 
activities recognize and initiate transcription at the major 



Figure 4-7. Run-off transcriptional activity of the porcine 
and bovine RNA polymerase activity on the porcine D-loop 
template . Panel A. Effect of MgCl 2 and MnCl 2 on the 
ability of the porcine transcription extract to transcribe 
its homologous DNA template (pBB21 linearized at the Pst 1 
site). Varying amounts of MgCl 2 (lanes 2-6) and MnCl 2 
(lanes 7-11) were included in each reaction (0 in lane 1, 1 
mM in lanes 2 and 7, 2.5 mM in lanes 3 and 8, 5 mM in lanes 
4 and 9, 10 mM in lanes 5 and 10, 20 mM in lanes 6 and 11) 
and the products electrophoresed on a sequencing gel. Panel 
B. Run-off transcriptional activity of the bovine 
mitochondrial RNA polymerase on the heterologous porcine 
template. As above, the bovine transcription extract 
(fraction 7 from Figure 4-3) was incubated with the linear 
pBB21 template in run-off transcription reactions. In lane 
1, 4 ul of the bovine activity was added, 8 ul in lane 2. 



A. B. 

PIG COW 

M123456789 10 11 1 2 



95 









*— «. 



Figure 4-8. Primer extension of in vitro transcription 
reaction products using the bovine and porcine RNA 
polymerase activities on the porcine D-loop template . Panel 
A. Primer extension of in vitro transcripts. The porcine 
(lanes 1 and 3) and bovine (lanes 2 and 4) transcription 
extracts were incubated with undigested pBB21 in 
transcription reactions and annealed to a radiolabeled 
oligonucleotide primer (corresponding to 1054-1072 on the 
porcine mitDNA sequence) . Following extension with reverse 
transcriptase the reaction products were electrophoresed on 
a sequencing gel. Lanes 3 and 4 represent reactions that 
were pretreated with RNAse Tl prior to annealing of the 
oligonucleotide primer. Lanes A, G, C, and T represent 
dideoxy sequencing reactions of pBB21 using the same primer 
as above. Panel B. Sites of in vivo and in vitro 
transcription initiation. Sequences from pBB21, near the 
porcine L-strand RNA start site, are shown. The underlined 
region indicates an 11 bp sequence seen in the clone, but 
not in the in vivo porcine RNAs analyzed in Figure 3-4. 
Sites of in vivo and in vitro transcription initiation 
within this region are indicated by heavy and light arrows, 
respectively. 






A. 



97 



B. 



A 


G 


C 


T 


^^^^^4 


^3 














hSm 















* 



IN VIVO 

CACAAACCACCATATQTTATMMCACTTATAAAACACTTACTTAAATACGTGCTACGAA 
IN VITRO T A A A 









98 

L-strand RNA start site, generating a triplet of labeled 
species centered at this location. We also note another 5' 
end downstream which corresponds to the smaller prominent 
labeled species seen in the run-off transcription experiments. 
In some but not all preparations of in vivo mitochondrial RNA, 
we were also able to detect minor RNA 5 ■ ends which map to 
this position, suggesting that the in vitro band seen at the 
same position may not be artif actual. The appearence of this 
band may be due to the 10 bp insertion seen in the cloned 
porcine mitochondrial DNA (shown underlined in Figure 4-8, 
Panel B) relative to the RNAs used to determine the L-strand 
RNA 5' end shown in Figure 3-4. Other, less prominent, RNAs 
were also detected in both the run-off and primer extension 
assays. 

Discussion 
We have isolated RNA polymerase activities from the 
mitochondria of the pig and cow (two related species from the 
Artiodactyl order) capable of specific in vitro transcription 
on their respective cloned D-loop DNA templates. In all 
experiments to date, the bovine and porcine mitochondrial RNA 
polymerase activities were functionally indistinguishable, 
both transcribing DNA templates containing cloned bovine and 
porcine D-loop seguences as well as a synthetic Artiodactyl 
LSP in an identical manner. While both polymerase 
preparations were capable of specific initiation of RNA 






99 

synthesis on Artiodactyl templates, the relative preference 
of transcription initiation at certain sites appeared to be 
slightly variant from that found in vivo . When incubated with 
a cloned template containing the 5 ■ end of the bovine D-loop 
region, although both polymerase activities initiated 
transcription at the same site as the major in vivo L-strand 
RNA, they favored transcription from a position 19 bp 
upstream. On a related synthetic DNA template which contains 
sequences very similar to the bovine LSP region but has 
restriction enzyme sites separating domains conserved in 
Artiodactyl species, the primary site of transcription 
initiation also occurred at this -19 site. While both 
activities transcribed efficiently from the major porcine L- 
strand RNA start site, they also initiated with about equal 
efficiency from a position 30 base pairs downstream. 

Variant Transcription Start Site Selection In Vitro 

Although both bovine and porcine RNA polymerase 
activities were capable of recognizing the major bovine L- 
strand initiation site (nucleotide 241 and 242 on the bovine 
sequence) , the primary site of in vitro initiation from the 
LSP region occurred 19 nt upstream, a position corresponding 
to a relatively minor in vivo RNA 5' end. These results 
indicate that while our partially purified extracts are 
capable of accurate initiation, they have a somewhat variant 
preference for initiation on the bovine D-loop template in 



100 

vitro . A comparison of the sequences immediately surrounding 
the two start sites shows that these regions have limited, but 
discernable similarity, both containing a C residue about 3 
nt downstream from the initiation nucleotide, followed by an 
A/T rich region upstream (CTTAAATA, in vivo vs. CTTTTAAC in 
vitro ; initiation nucleotides are underlined) . Several 
factors could account for the differences seen between RNA 
synthesis in vitro and in vivo . First, an important protein 
or factor may be missing from the RNA polymerase preparations. 
However, the ability of both the porcine and bovine RNA 
polymerase activities to transcribe reasonably efficiently 
from the major porcine in vivo L-strand start site, suggests 
that all the necessary components are present in both 
preparations. Furthermore, investigations in other mammalian 
systems have thus far identified only two activities involved 
in in vitro transcription, an RNA polymerase and a DNA binding 
protein necessary for regulating efficiency of transcription 
(Fisher and Clayton 1985; Fisher and Clayton 1988; Bogenhagen 
and Insdorf 1988) . We can identify both of these activities 
in our extracts (see Chapter 5) . Detailed purification 
strategies in other laboratories have yet to identify any 
other proteins involved in this process. This does not rule 
out the possibility of an accessory protein as yet 
unidentified or unique to the bovine system, although the 
emergence of such a novel transcription factor in only 52 



101 
million years or so of evolutionary divergence from the pig 
seems highly unlikely. 

Another possibility which would explain the in vitro site 
selection is that although current studies suggest that only 
two proteins are involved in mitochondrial transcription, it 
is possible that the relative ratio of the two proteins in 
our fractions may be different from that found within the 
mitochondrion. A surplus of a transcription factor could 
conceivably lead to binding at secondary recognition DNA 
domains and thereby stimulate transcription at normally minor 
sites. Conversely, excess levels of RNA polymerase could have 
the same effect, resulting in transcription from alternate 
positions. It is this idea that will be expanded in Chapter 
6 to develop a unified model of mitochondrial transcriptional 
regulation. 

Another possibility is that we have identified a 
processed RNA 5' end in the bovine mitochondrial RNA, and that 
in vivo RNA synthesis indeed initiates 19 nucleotides upstream 
of that in vivo end. Although we cannot totally rule out this 
possibility either, the identification of a mitochondrial RNA 
of opposite polarity with its 5 1 end also at position 240 on 
the bovine seguence, as well as the seguence homology seen 
between porcine and bovine RNAs at the 5' ends of their 
respective L-strand transcripts, suggest that the majority of 
in vivo RNA synthesis initiates from the identified in vivo 
site. Additionally, no apparent function can be readily 



102 

ascribed to RNA processing 19 nucleotides downstream from a 
mitochondrial initiation site, and this type of activity is 
not seen in other vertebrate species, including the pig. 

A final and perhaps most appealing possibility takes into 
account the seguence dissimilarity seen between the pig and 
cow downstream from their respective major L-strand RNA 5' 
ends. The bovine sequence contains two runs of six or more 
C residues at this position while the porcine sequence is 
relatively A/T rich. The mitochondrial RNA polymerase in 
vitro may lack the ability to melt the two DNA strands at this 
site to initiate RNA synthesis due to the strength of the 
nearby C/G domain, preferring instead to initiate at a 
secondary site 19 bp upstream. In support of this idea is the 
relatively close proximity of the origin of the D-loop DNA in 
the cow, less than 60 bp downstream from the bovine L-strand 
RNA start site. The presence of this triple stranded region 
in vivo f only a few base pairs downstream from the LSP may 
create local changes in the DNA that partially destabilize the 
bovine C/G domain, allowing formation of an open initiation 
complex at the identified site. In contrast to the bovine 
system, the porcine L-strand start site is not G/C rich 
immediately downstream from the major L-strand RNA start site. 
This is consistent with its origin of heavy strand replication 
being positioned several hundred base pairs further downstream 
(Ghivizzani et al. submitted) Also consistent with this 
hypothesis are the presence of analogous runs of heavy strand 






103 

Cs in the water buffalo and giraffe mitDNAs, two additional 
species where CSB-1 and the putative D-loop DNA 5' ends are 
proximal to the L-strand RNA start sites. In support of this 
hypothesis, as will be demonstrated in Chapter 6, deletion of 
three of the 5 Cs within the bovine C/G rich domain resulted 
in the initiation of in vitro transcription preferentially at 
the major in vivo L-strand start site. 

Increased Levels of Nuclease Activity in Artiodactyl 
Mitochondria Obtained from Tissue 

Previous investigations in other laboratories using cells 

grown in culture had found success in isolating mitochondrial 

activities capable of specific in vitro transcription using 

relatively direct methods, sometimes finding that just cleared 

mitochondrial lysates were suitable for transcription 

experiments (Walberg and Clayton 1983; Chang and Clayton 1986; 

Bogenhagen and Yoza 1986a) . Unlike these reports, 

mitochondria purified from Artiodactyl tissue contained 

significant levels of RNA and DNA nuclease activity. One such 

activity, a site specific DNA endonuclease, often 

cofractionated with the RNA polymerase activity during 

chromatography of both bovine and porcine mitochondrial 

proteins. A DNAse activity of similar specificity has also 

been identified in bovine mitochondrial lysates from tissue 

culture cells (Low et al. 1987) which digests DNA at or near 

stretches of poly C residues. As shown in Figure 4-1, high 

levels of RNA nuclease activity were also present in protein 



104 
extracts from both cow and pig. When present at low levels 
(either as a result of fractionation from other RNAse 
activities or from dilution of the activity itself) in the RNA 
polymerase preparations this RNAse was also reasonably site 
specific, digesting in vitro transcription products from the 
porcine template near CSB-2 (data not shown) . This type of 
activity is most likely responsible for the RNA 5' ends which 
map to the bovine CSB-2+3 region in the cow as seen in Figure 
4-6, lane 6, upon adding increased levels of the bovine RNA 
polymerase activity to the transcription reaction. A 
mitochondrial RNA nuclease with analogous activity has been 
identified in murine mitochondria (Chang and Clayton 1987) . 
It specifically digests RNA molecules between CSB-2 and CSB- 
3 guided by the RNA component of the RNP (Bennet and Clayton 
1990) . Although the precise biological function of this 
activity has not been determined, it has been proposed to be 
involved in creating priming sites for DNA replication (Chang 
and Clayton 1987) . The relative abundance of these nucleases 
in both the porcine and bovine tissue mitochondrial lysates 
compared to mitochondrial lysates from human and murine tumor 
cell lines may reflect inherent protein composition 
differences between mitochondria in these distinct cell types. 

Why Are Bovine and Porcine Mitochondrial Polymerase Activities 
Functionally Identical ? 

On all DNA templates tested, the in vitro transcriptional 

activities of the bovine and porcine RNA polymerase 



105 
preparations were indistinguishable, with both extracts 
initiating transcription at exactly the same sites with very 
similar efficiencies. This is the first report of such 
functional similarity between mitochondrial RNA polymerase 
activities from two different species. This functional 
similarity suggests that the proteins involved in porcine and 
bovine mitochondrial transcription must also be very similar 
and not rapidly evolving in order to accommodate the mutation 
rate of mitDNA in the regulatory domains of the D-loop. The 
inference, therefore, is that seguences conserved within the 
D-loop regulatory regions of these species are important for 
function and allow efficient heterologous in vitro 
transcription. The comparison of DNA sequences found at 
putative Artiodactyl LSP regions shows that the site of 
transcription initiation is the only major site of absolute 
sequence conservation and thereby suggests that this is the 
only domain where sequence recognition is absolutely critical 
to promoter function. A corollary to this is that if 
mitochondrial transcription factors interact within upstream 
regions of Artiodactyl LSPs, they must not be rigidly limited 
to fixed recognition sequences but are probably able to 
identify a wider range of DNA sequences. In order to further 
address these matters it will be necessary to identify 
transcription factors within fractions containing 
transcriptional activity and characterize their DNA 



106 
interaction with a series of homologous and heterologous 
promoter containing DNAs. This is the subject of Chapter 5. 



CHAPTER 5 

ISOLATION OF A BOVINE mitTF ACTIVITY AND 

CHARACTERIZATION OF ITS DNA-BINDING SPECIFICITY 



Introduction 
The mitochondrial transcriptional machinery has been 
demonstrated to contain at least two distinct components: an 
RNA polymerase activity and a transcription factor necessary 
for efficient transcription of the light and heavy strand 
promoters (Fisher and Clayton 1985; Bogenhagen and Insdorf 
1988) . In the human and mouse, mitochondrial transcription 
factors bind DNA seguences between 12 and 40 base pairs 
upstream from transcriptional start sites (Fisher et al. 
1987; Fisher et al. 1989). Factor binding at these 
recognition seguences correlates with activation of specific 
in vitro transcription at the proximal downstream start site 
by the mitRNA polymerase (Fisher et al. 1987). In both 
these species the strongest binding occurs at the light 
strand promoter. This preferential LSP binding directly 
correlates with the relative strengths of the two promoters 
(LSP and HSP) in vitro , with light strand transcription 
occurring significantly more efficiently than heavy strand 
transcription (Chang and Clayton 1984; Chang and Clayton 
1986a; Fisher et al. 1987). The mitochondrial transcription 

107 



108 
factor (mitochondrial transcription factor 1: mitTF 1 
(Fisher and Clayton 1988)) responsible for this binding in 
humans has been recently cloned and sequenced (Parisi and 
Clayton 1991) . It is a 25 kilodalton protein which has 
distinct but limited similarity to several types of high 
mobility group (HMG) proteins. HMG proteins have been shown 
to be involved in a wide variety of relatively nonspecific 
protein :DNA interactions including chromatin assembly and 
transcriptional activation. 

While the human and mouse mitochondrial RNA polymerase 
activities appear restricted to transcription of their 
homologous D-loop templates, their respective mitochondrial 
transcription factors can be functionally exchanged in 
heterologous DNA binding and transcription assays (Fisher et 
al. 1989). These transcription factors are able to bind 
specifically at both mouse and human LSP recognition 
sequences, and confer specific transcriptional activity to 
the mitochondrial RNA polymerase of either species (Fisher 
et al. 1989) . This functional interchangeablity is rather 
paradoxical in the sense that each transcription factor 
interacts at the same region on both human and mouse 
template mitDNAs despite an apparent lack of sequence 
similarity between these LSP regions. 

Previously, we determined the sites of light strand 
transcription initiation in several more closely related 
species within the Artiodactyl order (cow, water buffalo, 



109 
giraffe, and pig) (Chapter 3) . A comparison of their 
sequences immediately upstream showed that several blocks of 
sequence had a limited evolutionary conservation, suggesting 
their importance as transcriptional regulatory domains. In 
addition, one of the species analyzed (pig) appeared to have 
a 20 bp insertion which separated the conserved domains from 
the downstream RNA start site. When mitochondrial extracts 
containing RNA polymerase activity from the pig and cow (the 
two most diverse Artiodactyl species studied) were used to 
transcribe their homologous and heterologous DNA templates 
in vitro , both had identical transcriptional activities on 
either template (Chapter 4) suggesting that the proteins 
involved in RNA synthesis are highly conserved but that they 
have relaxed sequence specificities. 

In an attempt to more clearly delineate the sequence 
requirements involved in mitochondrial promoter regions and 
to better understand the apparent relaxed sequence 
requirements for in vitro mitochondrial transcription, we 
have isolated protein fractions from bovine tissue which 
contain mitochondrial transcription factor activity but no 
RNA polymerase activity. We have tested the ability of this 
activity to recognize and bind specifically to a range of 
evolutionally diverse D-loop DNA sequences. Similar to 
other mammalian systems, the bovine factor recognizes and 
binds to sequences about 12 to 4 bases upstream of both its 
light and heavy strand RNA start sites. In heterologous 



110 
binding experiments, the bovine factor footprints a 
relatively larger region upstream from the porcine light 
strand RNA start site from about -13 to -60. Contained 
within the protected region is the 20 base pair domain whose 
seguence dissimilarity to other Artiodactyl LSP regions 
suggested its presence as a DNA insertion. The bovine 
factor is also able to bind seguences upstream from the 
light and heavy strand RNA start sites in the human and 
mouse, as well as recognize a range of seguences clearly 
outside the promoter regions of these species. A comparison 
of all mitochondrial promoter seguences recognized by the 
bovine factor demonstrates that sites of transcription 
factor binding consist not of specific seguences but are 
better described as seguence motifs. These relaxed seguence 
recognition reguirements further suggest that transcription 
factor binding may not be directed solely to regions within 
the mitochondrial D-loop. 

Results 
Fractionation of protein extracts over phosphocellulose 
has previously been used to separate mitochondrial 
transcription factor and RNA polymerase activities for both 
the human and mouse systems (Fisher and Clayton 1985; Fisher 
et al. 1989) . In our laboratory, the separation of bovine 
mitochondrial proteins over phosphocellulose resulted in the 
cofractionation of DNA nuclease activity with the RNA 



Ill 

polymerase, making this type of protein purification 
unsuitable for our needs. Previously, however, we had 
success isolating mitochondrial protein extracts capable of 
in vitro transcription by using a KC1 step gradient to elute 
bound proteins from DEAE-Sephacel. In an effort to resolve 
bovine transcription factor activity from RNA polymerase for 
use in protein-DNA interaction studies, we again bound 
mitochondrial extracts to DEAE-Sephacel, but changed our 
elution profile to a relatively shallow linear gradient 
(0.05 to 0.5 M KCL) . As in Chapter 4, mitochondrial lysates 
were prepared and loaded onto a 60 ml column of heparin- 
Sepharose at 0.2 M KCL, and bound proteins batch eluted with 
a 0.6 M KC1 step gradient. Column fractions containing 
eluted proteins (as determined by spectrophotometric 
absorbance at 280 nm) were collected and pooled. Column 
buffer was then added in sufficient volume to dilute the KC1 
concentration of the pooled fractions to 0.05 M (verified by 
conductivity) . The diluted mixture was then loaded onto a 
50 ml column of DEAE-Sephacel, and bound proteins were 
eluted using a 40 ml linear gradient from 0.05 to 0.5 M KCl. 
Protein fractions of 3 mis were collected, dialyzed 
individually, and assayed for non-specific RNA polymerase 
activity on denatured herring sperm template DNA. 

As shown in Figure 5-1, Panel A, nonspecific RNA 
polymerase activity was detected in several fractions, 
eluting from the column at 0.15 M to 0.375 M KCL in 



112 
fractions 9-15, with peak activity in fraction 12. The 
fractions were then tested for specific transcriptional 
ability on the synthetic Artiodactyl D-loop template in a 
series of run-off transcription reactions. As seen in 
Figure 5-1, Panel B only fractions 9-12 were able to 
generate detectable transcripts of the appropriate size. 
These fractions, however, represented only about the first 
half of the total RNA polymerase activity; those in the 
second half of the peak (fractions 13-15) , which eluted at 
higher KC1 concentrations had no detectable specific 
transcriptional ability although substantial RNA polymerase 
activity remained. 

As reported by Fisher and Clayton (1987) stimulation of 
RNA synthesis involves specific sequence recognition and 
binding of promoter sequences by the transcription factor 
mitTF 1. To determine candidate fractions from which to 
test for bovine mitochondrial transcription factor activity, 
we analyzed eluted proteins for non-specific DNA binding 
using a band-shift assay. A radiolabeled DNA fragment 
containing the synthetic Artiodactyl light strand promoter 
sequence was incubated in binding reactions with each of the 
protein fractions from the DEAE-Sephacel column. As seen in 
Figure 5-2, Panel A several fractions from the early part of 
the gradient (fractions 6 through 12) were able to retard 
the migration of the labeled fragment in the gel. All 
detectable DNA binding activity eluted from the column 



Figure 5-1. RNA polymerase activity of bovine mitochondrial 
proteins eluted from DEAE-Sephacel by linear gradient . 
Panel A. Non-specific RNA polymerase activity in fractions 
6-18. Bovine mitochondrial proteins, batch eluted from a 
heparin-Sepharose column, were bound to DEAE-sephacel and 
eluted by a linear gradient (described in Materials and 
Methods) . Collected fractions were assayed for their 
ability to incorporate radiolabeled nucleoside triphosphates 
into acid precipitable counts using denatured salmon sperm 
DNA as template. Panel B. Specific RNA polymerase 
activity. Fractions 6-18 were assayed for their ability to 
transcribe the pALSP template (linearized at the Sst 1 site) 
in run-off transcription reactions. The arrow designates 
the major in vitro transcript (-19) described in Chapter 4. 



114 



A. 



20000 -i 












Q 

I 

2 

o 

O 10000 



§ 




t — i — i — i — i — i — i — i — i — i — i — r 

10 15 

FRACTION NUMBER 



— I 
20 



B. 



6 7 8 9 10 11 12 13 14 15 16 17 18 




115 

between 0.07 M and 0.25 M KC1. We then switched to a 
DNAse 1 protection assay, in order to determine if we had 
isolated proteins capable of binding the bovine D-loop 
region in a seguence specific manner. As shown in Figure 5- 
2, Panel B we were able to identify several fractions from 
the early part of the column (fractions 7-12) which 
footprinted in two regions of the bovine D-loop, at 
positions upstream from both the light and heavy strand RNA 
start sites respectively. These fractions correlated 
directly with the band shift DNA binding activity. 

Also apparent in footprinting assays with fractions 
containing specific DNA binding proteins was the appearance 
a series of labeled bands which migrated with a run of poly 
G residues in the corresponding seguencing ladder. These 
bands did not represent regions of DNAse 1 hypersensitivity, 
but instead were the effects of a residual level of the 
mitochondrial DNAse activity described in Chapter 4. An 
identical pattern of labeled species was generated in this 
region in footprinting reactions in which no DNAse 1 was 
added (Figure 5-5, Panel A), confirming this point. 

When the profiles of fractions with specific DNA- 
binding and non-specific RNA polymerase activities were 
compared, only fractions where the DNA footprinting and RNA 
polymerase activities overlap (fractions 9-12) could 
specific RNA transcripts from the linearized template be 
generated (Figure 5-3) . From these results it seemed very 



Figure 5-2. DNA-binding activity of the bovine 
mitochondrial proteins eluted from DEAE-Sephacel bv linear 
gradient . Panel A. Non-specific DNA-binding activity. 
Fractions 6-18 were assayed for their ability to retard the 
migration of a radiolabeled, 130 bp, Kpn 1-Not 1, DNA 
fragment from pALSP in band shift gels. This DNA contained 
the Artiodactyl synthetic promoter region and approximately 
60 bp of Bluescript vector sequence. The arrow designates 
the migration of the unbound fragment. Panel B. Specific 
DNA-binding activity. Fractions 6-18 were assayed for the 
ability to specifically bind to a 592 bp, 32 P labeled, Bam 
Hl-Hinc II DNA fragment containing the bovine LSP and HSP 
region (corresponding to bases 16202-457 on the bovine 
mitDNA) in DNAse 1 protection assays. Lanes A, G, C, and T 
designate corresponding dideoxy sequencing reactions of the 
probe DNA. Transcription initiation sites and direction of 
RNA synthesis are designated, HSP and LSP. 



6 7 8 9 10 1112 13 14 15 16 17 18 



B 



LSP 



,1 




117 






\ 

HSPI 



1 








Figure 5-3. Comparison of non-specific RNA polymerase 
activity with specific DNA-binding and RNA polymerase 
activities . The profiles of fractions containing non- 
specific RNA polymerase activity and specific DNA binding 
ability are compared. From Figure 5-1, Panel A, fractions 
9-15 had detectable RNA polymerase activity. Fractions 6-12 
demonstrated the ability to specifically recognize and bind 
to sequences upstream from bovine mitochondrial 
transcription initiation sites (Figure 5-2). Fractions 
capable of specific transcription initiation (as determined 
in Figure 5-1, Panel B) are designated by the cross-hatched 
region. 



119 



20000 -i 



1.2 



f 



Q 

UJ 



< 

BE 
O 
a. 
oc 

O 10000 

o 



s 
a 
o 




I i — i — r 



- 1.0 



f 

> 

-0.8 H 



0.6 



-0.4 



0.2 



i - - - r - J- oo 

'■ | i i ■ r — i f 



20 



o 

< 

o 

z 

a 

z 

m 

< 
z 
a 



FRACTION NUMBER 



120 
probable that bovine mitTF activity was present in fractions 
9 through 12 , and represented at least a subset of the DNA 
binding proteins found in Figure 5-2. 

To confirm chromatographic separation of 
transcriptional components, we attempted to reconstitute in 
vitro transcriptional activity by the addition of DNA- 
binding proteins from the early part of the elution gradient 
to column fractions with only non-specific RNA polymerase 
activity, in the trailing (higher KCl) part of the gradient. 
Fraction 8 (containing only DNA binding activity) and 
fraction 13 (containing non-specific RNA polymerase 
activity) were assayed for specific transcription both 
separately and mixed together in run-off transcription 
reactions (Figure 5-4) . Individually neither fraction could 
initiate RNA synthesis from the light strand promoter region 
at a significant level. However, when aliquots of each 
fraction were mixed together, specific transcriptional 
activity was restored to the RNA polymerase activity. 
Similar to the human system, these results indicate we have 
identified protein fractions which contain bovine 
mitochondrial transcription factor activity and that bovine 
mitochondrial transcription also requires at least two 
elements for initiation in vitro : an RNA polymerase and an 
activity with specific DNA-binding and footprinting 
properties. 



Figure 5-4. Reconstitution of specific transcriptional 
activity from DNA binding and non-specific RNA polymerase 
activities . Panel A. Summary of activities contained in 
fractions 8, 10, and 13 (eluted from the DEAE-Sephacel 
column described in the text and depicted in Figures 5-1 
through 5-3). Panel B. Reconstitution of specific 
transcription activity. Lane 1 represents run-off products 
from incubation of 8 ul of fraction 10 with pALSP 
(linearized at the Sst 1 site) . Lanes 2 and 3 represent 
run-off products from incubation of 8 ul of fractions 8 and 
13, respectively, with the same template. Lane 4 represents 
transcription products when 4 ul of fractions 8 and 13 were 
mixed together in similar run-off transcription reactions. 
Lane M corresponds to radiolabeled fragments from a Hpa II 
digest of pBR322 used as DNA size standards. 



Fraction Specific DNA Non-Specific Specific 

Number Binding RNA Polymerase Transcription 



122 



8 



1 



1 3 



B. 



Fraction 
Number 


Microliters 


Added 




8 


— 


8 





4 


10 


8 


— 





— 


13 


— 


— 


8 


4 





m 



M 1 



123 
To precisely determine the sites of transcription 
factor interaction relative to the bovine light and heavy 
strand RNA start sites, we incubated the bovine 
mitochondrial transcription factor activity with DNA 
fragments containing LSP seguences labeled on either the 
light and heavy DNA strands in DNAse 1 protection assays. 
As shown in Figure 5-5, as increasing amounts of the 
fraction were added to the reaction, discreet areas of 
protection became apparent upstream from both the L- and H- 
strand RNA start sites (labeled LSP and HSP, respectively, 
in the figure) . In the LSP region, slightly offset 30 bp 
regions were protected on both DNA strands, corresponding to 
positions -14 to -45 (relative to the in vivo RNA start 
site) on the light strand and positions -15 to -46 on the 
heavy strand. Upstream from the H-strand RNA start site, 
less prominent regions of protection were also detected at 
about positions -9 to -4 and -12 to -4 from the H-strand 
start site on the light and heavy strands, respectively 
(diagrammed in Figure 5-10, Panels A and B) . 

Since the bovine and porcine RNA polymerase 
preparations both transcribed the porcine template in an 
identical fashion, we tested the ability of the bovine 
transcription factor to recognize and bind to seguences in 
the porcine LSP region. A porcine mitochondrial DNA 
fragment corresponding to bases 1302 to 1062, was end 
labeled on either the light and heavy strands and incubated 






Figure 5-5. DNA-binding activity of the bovine mitTF on its 
homologous D-loop template . Panels A and B represent the 
interaction of the bovine mitTF activity on the light (Panel 
A) and heavy (Panel B) DNA strands of the bovine D-loop. 
Either strand of the 592 bp, Bam Hl-Hinc II DNA fragment 
corresponding to nucleotides 16202-457 of the bovine 
sequence was radiolabeled near position 457 and incubated 
with various amounts of bovine mitTF activity in DNAse 1 
protection assays. Numbers at the top of each lane indicate 
the ul amounts of mitTF activity added in each assay. The 
far right-hand lane in Panel A had no DNAse 1 added after 
incubation of the binding proteins with the probe DNA. The 
boxes with solid lines designate regions of specific binding 
upstream of the L-strand RNA start site (labeled LSP) . 
Dashed boxes indicate weaker binding seen upstream of the H- 
strand RNA start site (as determined by King and Low 1987) . 
Sites of transcription initiation and direction of synthesis 
are indicated by arrows. Lanes A, G, C, and T represent 
corresponding dideoxy sequencing reactions of the DNA 
templates. Heavy lines adjacent to the sequencing reactions 
designate the region of the bovine mitDNA with sequence 
similarity to the proposed 20 bp insertion seen upstream 
from the porcine L-strand RNA start site (see Chapter 3) . 






A. 



AGCT01 24 808 




-- 







t 



LSP 



I 



B. 



125 



12 4 8 A G C T 



ISP 




I 



£.- 



HSP HEP 



" = - 



: - 




- 

I 






126 

with bovine transcription factor in heterologous 
DNAse 1 protection assays. As shown in Figure 5-6, the 
bovine factor interacted with relatively larger regions of 
DNA upstream from the porcine L-strand RNA start site, 
corresponding to positions -13 to -61 and -10 to -59 on the 
light and heavy strands, respectively. Sites of DNAse 1 
hypersensitivity were noted on the flanks of the protected 
region as well as within. Quite surprisingly, the most 
strongly protected region on both strands contains the 20 bp 
porcine insertion sequence (Chapter 3) , while the regions 
upstream from the insertion, which by sequence alignment 
appear to be most similar to other Artiodactyl LSP 
sequences, were primary sites of DNAse 1 hypersensitivity. 
Interestingly, the region of the bovine sequence most 
similar to the porcine insertion is located adjacent to the 
bovine phenylalanine tRNA gene. It served as a site for 
digestion by mitochondrial endonuclease and was not 
protected in any detectable manner (Figure 5-5, Panel A). A 
weak footprint was also noted immediately downstream from 
the porcine L-strand RNA start site followed by a region of 
DNAse 1 hypersensitivity near position +30. This region of 
hypersensitivity corresponds to a site of in vitro 
initiation of RNA synthesis by both bovine and porcine 
mitochondrial transcription systems (Chapter 4) . 

From the experiments above, we noted the ability of the 
bovine transcription factor to recognize and bind to 



Figure 5-6. Specific binding by the bovine mitTF to the LSP 
region of the porcine D-loop .. Similar to above, increasing 
amounts of bovine mitTF activity were incubated with DNAs 
containing the porcine D-loop region (corresponding to 
nucleotides 1062-1304) in DNAse 1 protection assays. The 
double stranded DNA fragments were labeled either on the 
light strand (Panel A) or heavy strand (Panel B) near 
nucleotide 1304. Regions of specific binding are indicated 
by open boxed regions. Darkened boxes designate sites of 
DNAse 1 hypersensitivity. L-strand transcription initiation 
sites and direction of RNA synthesis are indicated by arrows 
labeled LSP. Lanes A, G, C, and T represent corresponding 
dideoxy sequencing reactions of the porcine D-loop template. 
Heavy lines adjacent to the sequencing reactions designate 
the region upstream from the porcine L-strand RNA start site 
proposed to be an insertion relative to other Artiodactyl 
species (see Chapter 3). 






128 



A. B. 

AGCTO 124 80 AGCT 012480 




==_;::_-_: t z, 



LSP 







1 



I- 



129 

sequences upstream from both the bovine and porcine L-strand 
start sites. In an attempt to further delineate the limits 
of sequence recognition by bovine mitTF, we tested the 
ability of the factor to recognize and bind to D-loop DNAs 
from human and mouse. When incubated with a human D-loop 
region fragment, we found that the bovine factor protected 
the identical region bound by the human factor upstream from 
the L-strand RNA start site (Fisher et al. 1987), with 
similar regions of hypersensitivity (Figure 5-7) . Also 
similar to the human, at high levels of added bovine factor 
a diffuse pattern of protection was seen at the human HSP 
region. Downstream from the human LSP region a third region 
of protection was also seen. The dark bands adjacent to 
this area, though, were the result of mitochondrial 
endonuclease digestion at the human CSB-2 region (Chapter 4) 
and not DNAse 1 hypersensitivity caused by mitTF 1 binding. 
Specific binding of the bovine factor to the mouse D- 
loop region also resulted in protection of sequences 
upstream from the murine L-strand RNA start site in a manner 
similar to that of the murine transcription factor (Fisher 
et al. 1989) (Figure 5-8) . The bovine protein also bound to 
a region upstream from the murine H-strand RNA start site. 
To date however, this HSP region has not been shown to be a 
binding region for the homologous mouse protein. As also 
seen in human and porcine DNAs, the bovine factor was able 
to protect specific sequences outside the established 






Figure 5-7. Specific binding of the bovine mitTF activity 
to the human D-ioop region . As above, increasing amounts of 
bovine mitTF activity were incubated with DNA fragments 
containing the human LSP and HSP regions (corresponding to 
nucleotides 323-655 of the human mitDNA sequence) . The DNA 
fragments were radiolabeled on the light strand near 
position 655. Regions of specific binding by the bovine 
mitTF are indicated by empty boxes. Boxes defined by solid 
lines and dashed lines indicate regions of strong and weak 
binding, respectively. Sites of binding by the human mitTF 
(Fisher et al. 1987) are indicated by the filled boxes 
labeled HUMAN. A site of nuclease digestion by the bovine 
extract near the human CSB-2 element is indicated by an 
asterisk. Sites of transcription initiation and direction 
of RNA synthesis are indicated by arrows. Lanes A and C 
indicate corresponding dideoxy sequencing reactions used as 
size markers. 



131 



A C 1 2 4 8 




: 


■»_ 


. "V- 


■ -4 || 







H 
U 
M 
A 
N 



I 



HSP 












Figure 5-8. Specific binding bv the bovine m itTF at the D- 
loop promoter regions of the mouse . Increasing amounts of 
bovine mitTF activity were incubated with radiolabeled DNA 
fragments containing the LSP and HSP regions of the mouse 
(corresponding to nucleotides 15973-9 on the murine mitDNA 
seguence) in DNAse 1 protection assays. The DNA fragments 
were radiolabeled on the light strand near nucleotide 15973. 
Sites of specific binding by the bovine mitTF are indicated 
by empty boxes, with regions of strong binding outlined with 
solid lines and regions of weak binding with dashed lines. 
The box shaded with dots, labeled MOUSE, corresponds to the 
region bound by the homologous murine mitTF (Fisher et al. 
1989) . CSBs-2 and -3 are as labeled, indicated by patterned 
boxes. Sites of transcription initiation and directions of 
RNA synthesis are indicated by arrows. Lanes A, G, C, and T 
are dideoxy seguencing reactions performed on the 
corresponding murine D-loop DNA. 


















AGCT01 2480 




133 



HSP 








M 


'*'••:' 


O 


:\':\ 


U 




s 




E 



I 



LSP 



11 


*•-* 


==~=l 


: 


— *2 _ -, mm , - . 


— 


. — .— BKit 




= *.W|-» 


— 


= . . •-. 




__ l_ ^-nB- - 


- — 


Z~= 


S 



* 



i>*>] 



CSB-2 



B33& 






CSB-1 



134 
transcriptional control region. A striking example is the 
solidly protected region between the mouse CSB-1 and -2 
elements. 

Because one major goal of this research project was to 
determine the role that sequences upstream from 
transcriptional start sites play in regulating RNA 
synthesis, we have synthesized and cloned a composite 
Artiodactyl light strand promoter region that contains 
blocks of sequence conserved upstream from L-strand RNA 
start sites in Artiodactyl species. For ease of 
manipulation each block has been flanked by restriction 
enzyme sites. These sites replace unconserved sequences and 
divide the putative regulatory domains into movable 
cassettes, allowing the formation of a variety of templates 
for use in transcription and protein :DNA interaction 
studies. As discussed previously, in vitro transcription by 
the bovine RNA polymerase/mitTF complex initiates at similar 
positions on both the native and synthetic templates. In 
order to confirm that the bovine mitTF also recognized and 
bound the synthetic template in a manner similar to its 
native DNA target, we performed a series of DNAse 1 
protection assays on this cassette template. As shown in 
Figure 5-9 in spite of the sequence variation caused by the 
restriction enzyme recognition sequences, the bovine 
transcription factor also bound specifically to the 
synthetic promoter in a manner virtually identical to that 



Figure 5-9. Specific binding by the bovine mitTF on the 
synthetic Artiodactyl LSP region . Panel A. Specific 
binding at the synthetic Artiodactyl LSP. As above, 
increasing amounts of bovine mitTF activity were incubated 
with a radiolabeled DNA fragment containing the Artiodactyl 
LSP region in DNAse 1 protection assays. The DNA used was a 
130 bp, Kpnl-Not 1 fragment of the pALSP clone, radiolabeled 
at the Not 1 site (approximately 40 bp downstream from the 
major in vivo L-strand start site) Conserved Artiodactyl 
domains found near the site of L-strand transcription 
initiation are indicated by the boxed P, A, B, and C+D 
regions. In both Panels A and B, the major in vivo and in 
vitro sites of transcription initiation and directions of 
RNA synthesis are designated by the solid and dashed arrows 
respectively. Panel B. Specific seguences at the 
Artiodactyl LSP region bound by the bovine mitTF activity. 
Protected regions are designated by solid lines. Sites of 
DNAse 1 hypersensitivity are indicated by dots. 



136 



B 



A. 



2 4 8 




c 

+ 
D 



B -% 




AATTCCCCCTTAAATATCGATCCACCAAQCTTTTAACAOATCTTTTCCCTAO 

P ~A I B 



ATACTTATTTAAATTTTTCACG 



C+D 



3 



137 
of the native bovine LSP sequence. This demonstrates that 
bovine mitTF:DNA recognition must reside 

within the unaltered conserved domains in this synthetic 
LSP, or at least is tolerant of sequence variability at the 
restriction enzyme digestion sites. 

Discussion 
By changing the method of eluting mitochondrial 
proteins bound to a DEAE-Sephacel column from a step 
gradient to a linear gradient, we have been able to 
partially resolve two activities whose simultaneous presence 
is required for specific in vitro transcription initiation 
on the bovine mitochondrial genome. The ability to 
reconstitute specific transcriptional activity by the 
addition of two protein fractions, one containing only DNA 
binding activity and the other only non-specific RNA 
polymerase, demonstrates that like all other mitochondrial 
transcription systems studied (human, mouse, Xenopus, and 
yeast) specific RNA synthesis in bovine mitochondria 
requires at least two separable components. While it is 
conceivable that the specific DNA footprinting activity 
which fractionated in the early part of the gradient is 
unrelated to whatever component confers specificity to the 
RNA polymerase activity, the correlation of specific DNA 
binding upstream from sites of in vivo transcription 
initiation with the ability to transcriptionally activate 



138 

the RNA polymerase offers strong evidence to the contrary. 
Additionally, the human mitTF, which has been purified to 
homogeneity (Fisher and Clayton 1988) , has similar in vitro 
properties on its homologous DNA template (Fisher and 
Clayton 1985; Fisher et al. 1987; Fisher et al. 1989). 

Although we were able to reconstitute specific 
transcriptional activity by the addition of the binding 
activity to the RNA polymerase, the mitochondrial extracts 
isolated were not suitable to study the effects that 
variations of the relative concentrations of the two 
components may have on in vitro transcription. This was 
partially due to the presence of a residual level of 
mitochondrial DNA nuclease activity which cofractionated 
with the binding activity in the early part of the gradient 
(Figure 5-2, panel A). Addition of large amounts of 
transcription factor activity therefore resulted in a 
concomitant loss of intact DNA template and prevented 
reliable quantitation of transcriptional activity. Also, we 
found that RNA polymerase activity once depleted of 
transcription factor was considerably less stable under the 
storage conditions employed, with a continual loss of 
activity occurring upon the onset of isolation. 



The Bovine mitTF Binds Specifically at Artiodactvl Promoter 
Sequences 



The bovine mitochondrial transcription factor 



139 
recognizes and binds specifically to its homologous LSP and 
HSP regions. Binding at these sites occurs on both DNA 
strands at positions about -15 to -45 relative to the in 
vivo RNA start site. Protection from DNAse 1 digestion 
appears to be stronger at the LSP region, where discreet 
sites of protection as well as regions of DNAse 1 
hypersensitivity are seen. Upstream from the H-strand RNA 
start site, DNAse 1 protection, although detectable, is less 
defined with no apparent sites of hypersensitivity. A 
comparison of the core sequences protected at both sites 
shows the existence of limited sequence conservation (Figure 
5-10, Panels A, B, and D) . Generally, both sequences are 
A/T rich, contain few G residues and have similarly spaced 
runs of T residues. Sites of significant sequence 
similarity are found toward the 5' and 3' ends of the 
protected regions. Although transcription from the H- and 
L-strand start sites proceeds in opposite directions their 
transcription factor recognition sequences have the same 
polarity (discussed further below) . 

Specific binding by the bovine factor was also seen 
upstream from the porcine L-strand transcription start site. 
Although protein: DNA interaction was seen on both DNA 
strands at positions -35 to -60 (sites of limited similarity 
with other Artiodactyl LSP regions) , the region of strongest 
DNAse 1 protection occurred in a stretch of sequence which, 
due to its dissimilarity to other Artiodactyl sequences, was 



140 
suggested to be an insertion event (Chapter 3) (Figure 5-10, 
Panels C and D) . Interestingly, the analogue to this 
sequence in the bovine D-loop (located immediately upstream 
from the phenylalanine tRNA gene) was not protected from 
DNAse 1 digestion in similar assays, but instead was a 
preferred site of cleavage by a mitochondrial endonuclease 
present in the DNA binding fraction. Thus, this sequence, 
in itself, does not define mitTF binding. Several 
explanations may account for this variant interaction by the 
transcription factor at two very similar sequences. The 
first possibility is that the four bases that are different 
in these two regions are in positions critical to 
recognition of the sequence by the factor. However, the 
sequence dissimilarity found between the protected sequences 
at the bovine and porcine LSP regions (as well as at the 
human and mouse promoter regions discussed below) suggests 
that the absence or presence of these specific bases has 
little effect on transcription factor binding. A second 
alternative is that specific binding at mitochondrial 
promoters is directed by sequences outside the region of 
DNAse 1 protection. This scenario seems unlikely since 
directed deletion mutagenesis of sequences within protected 
regions can eliminate specific binding at that site (Chapter 
6) . A more probable explanation is that sequences important 
to transcription factor recognition were formed by the 
junction of the 20 bp region at the site of insertion. An 



141 
alignment of the sequences protected at the porcine and 
bovine LSP regions shows that the 8 bp sequence, CCCTAGAT, 
is present in both DNAs at identical positions relative to 
their respective in vivo transcriptional start sites (Figure 
5-10, Panels A and C) . This sequence is formed by the 3 
terminal nucleotides of the 20 bp porcine insert and the 
next 5 nucleotides of the porcine sequence. As discussed 
below this 8 bp stretch of DNA contains a sequence domain 
conserved in promoter sites bound by the bovine 
transcription factor regardless of its species origin 
(Figure 5-10, Panel D) . 

The size of the region of protein: DNA interaction 
between the bovine mitTF and the porcine LSP region was 
about 50% greater than that seen at either the bovine LSP or 
HSP (45 bp and 3 bp at the porcine and bovine LSP regions 
respectively) . From the published amino acid sequence of 
the human mitTF 1 (Parisi and Clayton 1991) mitochondrial 
transcription factor proteins appear to have a DNA binding 
domain of discreet size. Therefore, this larger region of 
bovine protein .-porcine DNA interaction suggests that tandem 
binding of multiple protein molecules may occur at the 
porcine LSP. This possibility is supported by the 
occurrence of two distinct regions involved in the site of 
protein interaction, the 2 bp insertion sequence and the 
sequences immediately downstream with identifiable 
similarity to other Artiodactyl LSP regions. Since the 



142 
porcine binding region is not twice as large as the bovine 
footprint, perhaps a situation where three protein molecules 
are bound is most plausible. This would in turn require 
that the protected homologous regions in the bovine promoter 
regions as well as in human and mouse involve the binding of 
two protein molecules. An analogous situation has been 
documented for the mammalian transcription factor, AP-2 . 
AP-2 binds to its specific recognition sequences as a 
homodimer, generating rather large regions (50 bp) of 
specific protection in footprint assays (Williams and Tijan 
1991) . This example provides a precedent for the large 
footprints we see and the possibility that they are due to 
multiple mitTF interactions at a single recognition site. 

The Bovine Mitochondrial Transcription Factor Also Binds 
Specifically to D-loop Promoter Sequences of Unrelated 
Species 

Column fractions containing bovine mitTF activity were 
able to recognize and bind to specific sequences upstream 
from human and mouse D-loop promoter sequences. Within the 
LSP regions of each species, the bovine factor was able to 
identify the same sequences bound by the homologous human 
and murine mitTF. In addition, the bovine factor was able 
to bind to sequences upstream from the murine HSP region, a 
site not identified by its homologous transcription factor. 

Bovine mitTF activity was also able to footprint areas 
of DNAs downstream from LSP regions. The strongest 






Figure 5-10. Specific sequences bound by the bovine mitTF . 
Panel A. Specific sequences bound by the bovine mitTF at 
the bovine LSP region. In Panels A, B, and C light and 
heavy DNA strands are designated L- and H-, respectively. 
Solid lines indicate regions of DNAse 1 protection, while 
dots designate sites of DNAse 1 hypersensitivity. Arrows 
indicate major sites of in vivo transcription initiation and 
the direction of synthesis. The boxed sequences in Panels A 
and C show sites of sequence homology between the bovine and 
porcine LSP regions bound by the bovine mitTF. Panel B. 
Specific sequences bound at the bovine HSP region. Panel C. 
Specific sequences bound at the porcine LSP region. The 
dotted line highlights the proposed insertion in the porcine 
mitDNA relative to other Artiodactyl species (see Chapter 
3) . Panel D. Comprehensive alignment of all mitochondrial 
promoter regions bound specifically by the bovine mitTF. 
The sequences have been divided into three regions of 
moderate sequence similarity. To emphasize conserved 
sequence motifs, dashes in the figure were used to 
substitute for pyrimidine residues. All sequences shown are 
from the light strand in the 5' to 3 1 direction. Panel E. 
Sequence alignment of the region strongly bound by the 
bovine mitTF between the murine CSBs-2 and -3 elements with 
the bovine LSP region. Sites of sequence agreement are 
indicated with colons. Gaps indicate breaks in the 
sequences. 



144 



If CCCCCTTAAATATCTACCACCACTTTTAACAGACTTT'l CCCTAGAT ftCTTATTTAAA 



H- 



GGGGGAATTTATAGATGGTGGTGAAAATTGTCTGAAA/^3GGATCT^rGAATAAATTA 



B. 



L- TTTTTCACGCTTTCAATACTCAATTTAGCACTCCAAACAAAGTCAATATATAAAC 
H- AAAAAGTGCGAAAGTTATGAGTTAAATCGTGAGGTTTGTTTCAGTTATATAITTG 



• •••< 



L- CTTACTTAAATACGTGCTACGAAAGCAGGCACCTACCC CCCTAGA'I rTTTACGCCAATCTACCACAAATAAGTT 
H- GAATGAATTTATGCACGATGCffTCGTC'CGTGGATGG'c lGGGATCT/J AAAATGCGQTTAGATGGTGTTTATTCAA 



• ••- 



• »••• 



-••• 



-••••- 



-•••••- 



Human 


LSP 


CTTTTAACAG 


TCACCCCCCAAC 


TA ACAC 


Cow 


LSP 


CTTTTAACAG 


ACTTTTCCC 


TAGATAC 


Cow 


HSP 


CTTTCAATAC 


TCAATT 


TAG CAC 


Pig 


LSP 


CTACGAAAGC 


ACCTACACCTACCCCCC 


TAGATAT 


Mouse 


HSP 


TGACCAAAAC 


TTCTAATCATACTC 


TA TTAC 


Mouse 


LSP 


TTCCCAAAAT 


ATCACTTATATTT 


TAGCTAC 


Human 


LSP 


AA-AG 


—A AA- 


TA ACAC 


Cow 


LSP 


AA-AG 


A 


TAGATAC 


Cow 


HSP 


AA-A- 


— AA-- 


TAG CAC 


Pig 


LSP 


— A-GAAAG- 


A— A-A— A 


TAGATAT 


Mouse 


HSP 


-GA— AAAA- 


AA— A-A 


TA TTAC 


Mouse 


LSP 


AAAA- 


A— A A-A 


TAGCTAC 



E. 



Mouse CSB-1 
Cow LSP 



TACT CAATACCAAATTTTAACTCTCCAA 
CACTTTTAACA GACTTTT CCCTAGATAC 






145 
protection outside conventional promoter regions is seen on 
murine DNA between CSBs-1 and -2, two sequence domains 
believed to be important in processing LSP transcripts 
(Chang and Clayton 1987; Bennet and Clayton 1990). Specific 
DNA binding by a transcription factor at this site has no 
immediately apparent function since no detectable sites of 
in vivo transcription initiation occur in this region (Chang 
and Clayton 1985; Chang and Clayton 1986b). It is possible 
that the footprints found at non-promoter regions could be 
due to the presence of an unrelated mitochondrial DNA 
binding activity present in these fractions. However, this 
is most likely not the case since highly purified human 
transcription factor also identifies an analogous region on 
its homologous template (Fisher et al. 1987). These results 
provide direct evidence for the specific binding of mitTF 
activity outside of conventional D-loop promoter regions and 
suggest that specific binding of this nature may occur in 
numerous regions of the mitochondrial genome. 

Mitochondrial Transcription Factor Target Sequences Consist 
of Evolutionallv Conserved Motifs 

A comprehensive comparison of all mitochondrial 

promoter sequences protected by the bovine transcription 

factor is shown in Figure 5-10, Panel D. This alignment 

suggests a conserved pattern of organization for these 

regulatory regions which have accordingly been divided into 

three domains. The first domain consists of a five base 



146 
pair pyrimidine rich region followed by a several of A 
residues. The second domain consists of a larger pyrimidine 
rich block of variable size (7-15 bp) interspersed with A 
residues. The third domain is defined by the consensus 
seguence, TAGATAC. The existence of these domains becomes 
more apparent when pyrimidine residues are represented as 
dashes as seen in the figure. From this alignment of all 
footprinted promoter regions, the mitochondrial 
transcription factor appears to be not rigidly limited in 
its recognition of DNA sequences, but instead has a 
preference for sequences that are probably better described 
as motifs. The ability of the bovine mitochondrial RNA 
polymerase machinery to transcribe and specifically bind to 
the synthetic Artiodactyl promoter will enable us to test 
the relevance of these putative recognition domains, 
allowing us to correlate in vitro transcription with 
specific DNAse 1 protection at upstream sequences (Chapter 
6). 

These surprisingly relaxed requirements for bovine 
mitTF recognition sites suggest that specific binding and 
localization of the mitochondrial transcription factor is 
not unique to the D-loop region and may well occur in 
regions elsewhere in the mitochondrial genome. This 
hypothesis has been preliminarily tested in Chapter 7. 



147 

Recognition Sequences for mitTF Binding at HSP and LSP and 
the Direction of H- and L-Strand Transcription Are in 
Opposite Orientations 

Contrary to what might be expected, the sequences bound 

by the bovine factor at its LSP and HSP appear to have the 

same polarity because the best alignment of HSP and LSP 

sequences within the same species occurs on the same DNA 

strand (light strand DNAs are shown in Figure 5-10, Panel D 

and are compared in the same 5' -3' orientation). Although 

the same type of arrangement has been previously noted in 

the human system (Fisher et al. 1987) , its occurrence in 

Artiodactyl species as well as in the mouse indicates a 

functional conservation of the position of transcription 

factor binding with respect to L- and H-strand 

transcriptional start sites. The effect mitTF binding has 

on efficiency of transcription initiation at HSP versus LSP 

may be a fundamental distinction between types of promoter 

regions. That is, activation of transcription may be 

strongest when the RNA polymerase interacts with its cognate 

start site and specific polypeptide domains of the mitTF 

molecule. This in turn depends on the orientation of the 

mitTF recognition sequences relative to the transcriptional 

start site. 






CHAPTER 6 

INTERACTION OF BOVINE MITOCHONDRIAL RNA POLYMERASE AND 

TRANSCRIPTION FACTOR AT SITES OF TRANSCRIPTION INITIATION 



Introduction 
Mitochondrial transcription has been studied in a 
variety of organisms (human, mouse, Xenopus . and yeast) . 
Consistent among all species is the existence of at least 
two separable components required for specific transcription 
initiation in vitro : an RNA polymerase activity and a 
transcription factor required for initiation specificity and 
efficiency. Isolation of transcriptionally active protein 
extracts from the mitochondria of these species and analysis 
of their ability to transcribe wild type and mutant template 
DNAs in vitro, has shown significant differences in the 
structure of each respective transcriptional regulatory 
region. In yeast, a relatively small 11 bp sequence 
containing the initiation site has been shown to be 
sufficient for full transcriptional activity. Yeast RNA 
polymerase activity and specificity factor, together, form a 
complex that binds and initiates transcription from this 
region. Initial reports for Xenopus suggest a similar type 
of promoter structure, in which only an 18 bp sequence 

148 



149 
centered on the site of initiation serves as the minimal 
regulatory region for transcription. Although a 
transcription factor activity has been identified in Xenopus 
mitochondrial extracts, its role in transcriptional 
regulation has not yet been determined. In previously 
studied mammalian systems (human and mouse) a bipartite 
promoter has been demonstrated, in which regulatory regions 
appear to exist as one species specific sequence required 
for initiation and another upstream region from about -10 to 
-3 5 serving as a recognition site for the mammalian 
transcription factor (mitTF) . At the light strand promoter 
(LSP) region, mitTF binding has been shown to be necessary 
for efficient RNA synthesis at the downstream site by the 
RNA polymerase. Deletion mutagenesis of the transcription 
factor binding site is associated with a significant 
decrease in the efficiency of transcription initiation. The 
relationship between the mitochondrial transcription complex 
and the heavy strand promoter (HSP) region is less clear. 
Although the mitTF has been shown to interact with sequences 
upstream from the H-strand RNA start site, protein DNA 
interaction, here, is relatively weak in comparison to the 
LSP region. Deletion mutagenesis of the HSP mitTF binding 
region appears to have only a moderate effect on the 
efficiency of transcription initiation at the normal 
downstream RNA start site. A problem, therefore, exists in 
understanding precisely how the same mitTF and mitochondrial 



150 
RNA polymerase activities interact at different 
mitochondrial promoters and whether or not there are general 
rules for these interactions independent of species. 

To answer this question, we have isolated extracts from 
bovine and porcine mitochondria capable of specific in vitro 
transcription initiation. Despite having somewhat 
dissimilar LSP regions, the transcriptional activities of 
the bovine and porcine extracts were found to be 
functionally interchangeable. Isolation of the bovine mitTF 
and characterization of its interaction on D-loop DNAs from 
a variety of species suggests that localization of the 
factor on DNA is not rigidly sequence specific, but is based 
on recognition of sequence motifs encompassing a rather 
broad range of sequences. 

In this chapter, we describe the interaction of the 
bovine transcriptional apparatus on a synthetic DNA template 
which contains blocks of conserved sequence found upstream 
from the conserved L-strand RNA start site in Artiodactyl 
species. Each domain has a restriction site separating it 
from its neighboring domains, allowing for directed domain 
deletion analysis. By correlating specific transcriptional 
activity with mitTF interaction on a series of such deleted 
synthetic template DNAs, we have shown that binding by the 
mitTF at specific sites is not required for transcriptional 
stimulation at cognate start sites. Additional studies show 
that removal of C/G base pairs immediately downstream from 



151 
the in vivo L-strand RNA start site enables in vitro 
transcription to occur at the correct in vivo site. This 
may suggest a role for the mitochondrial D-loop itself in 
assisting transcription through local melting of DNA 
strands. Overall, these results indicate that mitochondrial 
transcription initiation is not tightly regulated at the DNA 
seguence level and that transcription initiation may not be 
limited to the D-loop region but may occur at numerous other 
sites in the genome. When considered in the context of 
experiments in human, mouse, and Xenopus , a unified model of 
transcriptional initiation is suggested, whereby the ratios 
of RNA polymerase to mitTF and mitDNA regulate a balance 
between transcription of coding regions and RNA synthesis 
which primes DNA replication. 

Results 
A 74 bp DNA oligonucleotide containing blocks of 
sequence evolutionally conserved among Artiodactyl species 
occuring at and upstream from L-strand transcription 
initiation sites, was synthesized and cloned into the 
Bluescript vector (pALSP, described in Figures 3-6 and 6-1) . 
For ease of manipulation, this oligonucleotide was 
engineered with the conserved domains (labeled P, A, B, and 
C+D) separated by restriction enzyme recognition sites, thus 
forming mobile cassettes. In previous experiments this 
synthetic LSP region has shown to be a faithful substitute 



152 
for the native bovine LSP sequence, both for in vitro 
transcription (Chapter 4) as well as for mitTF:DNA 
interaction (Chapter 5) . To determine minimum functional 
sequences, a series of mutant deleted LSP regions were 
generated using restriction enzyme sites, designed within 
the synthetic LSP and the Bluescript polylinker, as points 
of deletion and religation (described in Materials and 
Methods) . We then tested the ability of transcriptionally 
competent, bovine mitochondrial protein fractions (isolated 
as described in Chapters 4 and 5) to recognize and 
transcribe these templates in a series of footprint and 
transcriptional run-off reactions. Initially, we made a 
series of template DNAs in which sequence domains from the 
5' end of the synthetic LSP region were systematically 
removed (Figure 6-1) , and tested the ability of each to 
direct specific transcription. As shown in Figure 6-2, the 
intact pALSP (lane 1) and pRRALSP (lane 2) DNAs, when 
linearized at a downstream Sst 1 site within the Bluescript 
vector and incubated with the bovine mitochondrial 
transcriptional activity in run-off transcription reactions, 
directed specific initiation of RNA synthesis. The products 
were labeled transcripts which migrated at 76 nucleotides 
relative to the DNA size standards, corresponding to in 
vitro tanscription at a position -19 relative to the in vivo 
L-strand RNA start site (Chapter 4). Deletion of the 5' C+D 
domain (lane 3) appeared to have no significant effect on 



Figure 6-1. Diagram of deletion clones constructed from the 
Artiodactyl synthetic LSP . As described in Materials and 
Methods, a 74 bp double stranded DNA oligonucleotide 
containing sequence blocks (designated P, A, B, and C+D) 
conserved among Artiodactyls (shown in Figure 3-6) was cloned 
into the Bluescript vector (pALSP) . An aberrant DNA was also 
formed in the cloning procedure, where the oligonucleotide was 
ligated into the Eco Rl site only, resulting in the pRRALSP 
DNA. From these two DNAs, a series of deletion clones was 
constructed consisting of various combinations of the 
Artiodactyl LSP domains. Letters designate restriction enzyme 
sites: S=Sst 1, B=Bam HI, R=Eco Rl, C=Cla 1, H=Hind III, L=Bgl 
II, V=Eco RV, Sl=Sal 1. The dark boxes represent the T3 
promoter region contained within the Bluescript vector. 



pALSP: ,- 
S 



P A B C+D 



BR C II L 



SI 



T3 



pPAB: ,- 
S 



B R C 



P A B 

■ i i 1 — 

H H C SI 



T3 



pPCD: r- 
S 



B R 



I P | C + D 



SI 



T3 



pPA: 



r 



P A 



R C HC SI 



T3 



PP: r- 
S 



B R C 



T3 



Bluescript: r 



i iii — i 1 — 

B RVHC SI 



T3 






154 



pRRALSP: , , | p | A | B | C + P I 



BR CHI 



i i i r 

VHC SI 



T3 



Figure 6-2. In vitro transcription of a series of pALSP 5 1 
deletion clones . Cloned DNAs containing various combinations 
of the conserved Artiodactyl domains (pRRALSP in lane 1, pALSP 
in lane2, pPAB in lane 3, pPCD in lane 4, pPA in lane 5, pP 
in lane 6 and Bluescript vector in lane 7) were linearized at 
the Sst 1 site and incubated with the bovine transcriptional 
extract in a series of run-off transcription reactions. The 
reaction products were then electrophoresed on a sequencing 
gel. Lane M indicates radiolabeled DNA fragments from a Hpa 
II digest of pBR322 used as size marker. 






156 



M 




1 2 3 4 5 6 7 






mm 



m 




157 

transcription initiation at the -19 in vitro site, even 
though the deleted sequences represented approximately half 
of the region protected by the bovine mitTF (Chapter 5) . As 
expected, the removal of domains A and B (containing the 
major in vitro RNA start site and the downstream half of the 
mitTF protected region) from the intact LSP region resulted 
in a total loss of detectable specific transcription 
initiation (lane 4) . No detectable transcripts initiating 
from within Artiodactyl LSP regions were seen in any 
template after deletion of both B and C+D domains (lanes 5- 
7) . Removal of these regions, however, was associated with 
the appearance of a prominent in vitro labeled RNA species 
initiating within upstream vector sequences. Transcription 
of the Bluescript vector alone (also linearized at the Sst 1 
site, shown in lane 7) produced a labeled species which 
migrated at 105 nt, unequivocally demonstrating that this 
RNA was a product of initiation within vector sequences and 
proving that the bovine mitochondrial transcription extracts 
possess no absolute requirement for homologous mitDNA 
Artiodactyl domains. As discussed below, these vector DNA 
directed transcripts represent RNA synthesis from a site 
within the T3 promoter sequence which is located upstream 
from the Bluescript polylinker. 

In an attempt to correlate specific transcription 
initiation with bovine mitTF binding, the series of DNAs 
containing various combinations of Artiodactyl domains used 



158 
above, were tested for their ability to serve as recognition 
sequences for the binding of mitTF. The results of these 
experiments were then compared to the transcriptional data 
from Figure 6-2 and together are summarized in Figure 6-3. 
As shown previously, the intact region was bound by the 
factor 

primarily within the B and C+D domains. When domains A and 
B were removed from the region (template pPCD) , even though 
the LSP region no longer contained the major in vitro RNA 
initiation site (and lacked the ability to direct specific 
transcription) , specific mitTF binding was detected in the 
remaining P and C+D domains. On all DNAs with the C+D 
domain deleted, no significant specific binding was 
detected. While templates containing domain B (the major 
start site of in vitro transcription) were capable of 
directing specific RNA synthesis from within cloned 
Artiodactyl sequences, initiation at this site, however, was 
not dependent on strong upstream binding by the mitTF 
activity. 

An interesting feature of certain transcription 
experiments was the appearance of transcripts from a region 
within upstream vector sequences near the T3 promoter. As 
indicated in Figure 6-3, significant RNA synthesis from this 
region was only observed from templates in which domains B 
and C+D were deleted, regions which comprise the primary 
sites of in vitro transcription initiation and DNA-binding. 









Figure 6-3. Diagram of specific DNA binding and transcription 
by the bovine transcription extract on the cloned Artiodactvl 
LSP domains . Boxes labeled P, A, B, and C+D represent 
various combinations of cloned Artiodactyl LSP domains used 
in run-off transcription and DNAse 1 protection assays. Shown 
to the right is a summary of the bovine transcriptional 
machinery's ability to initiate transcription from the 
Artiodactyl LSP region (LSP) , recognize and bind specifically 
to the domains in footprint (FP) assays, and initiate 
transcription from the T3 promoter region contained within the 
cloning vector. 






160 




P 1 A | B | C + D 



I P I A |B| 

r~p~i i c+p 

i p i A I 

urn 

Bluescript 



LSP FP 

+ + + + 



T3 



+ 



+ + 



+ 
+ 

+ 









161 
As noted above and in Chapter 4, in vitro transcription 
of the bovine LSP region and related sequences resulted in a 
preferred site of initiation at -19 nt relative to the major 
in vivo RNA start site at +1. In order to determine the 
reason for this we tested the hypothesis that bovine 
sequences near the in vivo +1 start site were preventing the 
use of this site in vitro . Our primary concern was that the 
stretch of 5 C residues downstream from the in vivo RNA 
start site may be inhibiting in vitro initiation at that 
position by not allowing strand separation. We, therefore, 
synthesized the DNA oligonucleotide, 

GATCCTTAAATATCTACCACCACTTTTAACA (and its complement) , to 
replace the existing P, A, and B domains within the pALSP 
clone. This DNA contained the native bovine analogues of 
the P, A, and B domains but only two C residues immediately 
downstream from the in vivo start site (maintaining both C 
residues allowed for the integrity of the Bam HI cloning 
site) and the native bovine sequences upstream. Cloning 
this oligonucleotide in place of the original P, A, and B 
domains in pALSP resulted in the formation of a promoter 
sequence identical to the bovine LSP from position +3 
through position -58, with the exception of an extra T 
residue at position -27. Another template was made using 
the same DNA oligonucleotide, cloned into the Bluescript 
vector alone at the Bam HI site, thereby forming a deletion 
equivalent to the removal of the C+D domain (pCS-CD) . These 



162 
DNAs were incubated in in vitro transcription experiments 
and the 5' ends of RNA reaction products mapped by primer 
extension. In Figure 6-4, Panel A, lane 1, a predominant 
doublet of 5 • ends was seen which mapped to the same 
nucleotide position as the major bovine in vivo RNA start 
site at +1 (Figure 3-2) . As with the -19 start site, in 
vitro RNA synthesis at the major in vivo site was not 
dependent upon the presence of sequences residing in domain 
C+D. This is shown in lane 2 where in vitro transcription 
and primer extension of pCS-CD also generated the same 
doublet of bands at +1, with similar intensity. Therefore, 
removal of the homopolymer C domain immediately downstream 
of the in vivo RNA start site allowed correct in vitro 
initiation. 

In order to test the correlation of initiation of in 
vitro transcription at the major in vivo RNA start site with 
mitTF binding, we incubated the bovine mitTF activity with 
pCSl and pCS-CD in DNAse 1 protection assays. As expected, 
pCSl, which contained DNA sequences nearly identical to the 
bovine mitTF binding site was footprinted in a manner 
virtually identical to that of the native bovine LSP region 
(Figure 6-4, Panel C) . Similar to previous experiments with 
templates containing domains P, A, and B, transcription 
factor binding at the DNA containing bovine LSP sequences 
from position +3 to position -27 appeared to be very weak, 
detectable primarily by the appearance of DNAse 1 









Figure 6-4 . In vitro transcription of the bovine LSP region 
after deletion o f C residues downstream from the major in vivo 
L-strand RNA start site and the mitTF binding site . Panel A. 
In vitro transcription of pCSl and pCS-CD. Cloned DNA 
templates, pCSl (lane 1) and pCS-CD (lane 2), were incubated 
in transcription reactions with the bovine transcription 
extract. A radiolabeled oligonucleotide (universal -20 
pri.mer) was annealed to the RNA products and extended with 
reverse transcriptase. The reaction products were then 
electrophoresed on a seguencing gel. Arrows indicate the 
major sites of transcription initiation. Lanes A, G, C, and 
T are dideoxy seguencing ladders of the corresponding DNA 
seguence from pCSl. Panel B. Sites of in vitro transcription 
initiation on pALSP (top) and pCSl (bottom) . The arrow at the 
top seguence the indicates the major site of in vitro 
transcription initiation on bovine LSP templates containing 
runs of 5 or more C residues immediately downstream from the 
major i n vivo L-strand RNA start site (shown underlined) . The 
arrows at the bottom seguence show the major site of in vitro 
transcription initiation when 3 of the C residues immediately 
downstream from the major in vivo L-strand RNA start site 
(shown underlined) were deleted. Panels C and D. DNAse 1 
protection of the Artiodactyl domains in pCSl and pCS-CD. A 
110 bp fragment containing the synthetic bovine LSP region 
(domains P, A, B, and C+D from pCSl; see text) and a 74 bp 
fragment containing only domains P, A, and B (from pCS-CD) 
were incubated with the bovine mitTF activity in DNAse 1 
protection assays. Lanes designated, +, indicate assays which 
included 6 ul of a fraction containing mitTF activity. Lanes 
designated, -, indicate control assays which had no added 
protein fraction. Lanes C and T indicate corresponding 
dideoxy seguencing reactions of each DNA fragment. 






164 



B. 



r\ m 



G C T 1 2 




* 



CCCCCTTAAATATCTACCACCACTTTTAACAG 



tt 



C. D. 

C T - + 




C T - + 






i 




B 




gatCCTTAAATATCTACCACCACTTTTAACAG 



165 
hypersensitive sites at positions -22 and -24 (Figure 6-4, 
Panel D) . 

As mentioned above, run-off transcription experiments 
on DNA templates in which the in vitro transcription 
initiation site and mitTF binding region (domains B and C+D) 
had been deleted, resulted in the appearance of a single 
labeled species which initiated from a position upstream of 
the cloned oligonucleotide seguences, within the Bluescript 
vector. In order to determine the precise nucleotide 
seguence of this site of transcription initiation, we 
incubated the bovine transcriptional extract with the native 
Bluescript vector in an in vitro transcription reaction and 
primer extended the resultant RNA. Electrophoresis of the 
products alongside a corresponding dideoxy seguencing ladder 
showed that the bovine transcriptional apparatus initiated 
RNA synthesis within the T3 promoter region, generating a 
triplet of bands centered at a position 4 nucleotides 
downstream from the normal T3 RNA polymerase initiation site 
(Figure 6-5) . This surprising result suggested that mitTF 
and mitRNA polymerase may interact with the T3 promoter 
seguence in a manner very similar to the homologous T3 RNA 
polymerase. Footprint analysis of the interaction between 
bovine mitTF and this region showed the existence of weak 
DNAse 1 protection throughout and upstream of the T3 
promoter region (Figure 6-6) . Sites of DNAse 1 
hypersensitivity were seen flanking the protected region, 



Figure 6-5. In vitro transcription from the T3 promoter 
region by T3 RNA polymerase and the bovine transcription 
extract . Panel A. In vitro transcription of the T3 promoter 
region. The Bluescript vector was incubated in in vitro 
transcription reactions with the T3 RNA polymerase (lane 1) 
and the bovine transcription extract (lane 2) . A radiolabeled 
oligonucleotide primer (corresponding to bases 665 to 686 on 
the Bluescript vector sequence) was annealed to the RNA 
products and extended with reverse transcriptase. The 
reaction products were then electrophoresed on a sequencing 
gel. The solid arrow indicates T3 transcription products 
which initiated at the corresponding T3 start site. The 
dashed arrow indicates the 5' ends of transcripts synthesized 
by the bovine transcription extract. The boxed region 
indicates the T3 promoter region. Lanes A, G, C, and T 
indicate dideoxy sequencing reactions on the corresponding 
Bluescript DNA. Panel B. Nucleotide start sites of in vitro 
transcription. Sequences within the T3 promoter are shown and 
the sites of transcription initiation by the T3 RNA polymerase 
(heavy arrow) and the bovine transcription extract (dashed 
arrows) 



167 



A. 



A G C T 1 2 




B. 



13 



I 



CAQLl'iTlGITCCCTITAGTGAGGGTI^y^TrcOGAGCIT 

~, AAA 

BOVINE 
MIT. 



• i i 
■ i i 



Figure 6-6. Specific binding by the bovine mitTF activity at 
the T3 promoter region . A 3 07 bp, Not 1-Pvu II DNA fragment 
from the Bluescript vector was radiolabeled at the 5' end (at 
the Not 1 site at position 665) and incubated with increasing 
amounts of the bovine mitTF activity (as indicated) in DNAse 
1 protection assays. The dashed box designates a region of 
mitTF binding which extends into the T3 promoter region 
(indicated by the solid box) . The solid and dashed arrows 
indicate the sites of transcription initiation and direction 
of synthesis by the T3 RNA polymerase and bovine mitochondrial 
extract, respectively. Lanes A, G, C, and T indicate 
corresponding dideoxy seguencing reactions performed on the 
Bluescript vector. 



169 



A G C T 0124 8 



i 




F- 



I I 






==»ii : 



•••■■a 



170 
located immediately at, and 5 nucleotides upstream from the 
point of initiation of RNA synthesis by the bovine extract. 
The protected region began at a position -4 relative to the 
T3 initiation site and extended upstream to a region of 
DNAse 1 hypersensitivity at -3 0. It, therefore, appears 
that in the absence of its cognate promoter, the 
mitochondrial transcriptional apparatus can recognize a 
bacteriophage promoter reasonably faithfully. 

Discussion 
From a cloned synthetic Artiodactyl LSP region, we 
constructed a series of deleted mutant LSP regions 
consisting of different combinations of seguence domains 
conserved among Artiodactyl species, upstream from L-strand 
RNA start sites. These DNAs were used in a series of in 
vitro transcription and DNAse 1 protection analyses in an 
attempt to correlate transcription with mitTF:DNA 
interaction at seguences upstream from initiation sites, and 
to define minimal reguired cis- elements. The results 
demonstrate that relatively small seguences located at or 
near RNA start sites are capable of directing efficient in 
vitro transcription initiation, apparently reguiring only 
relatively weak interaction between the template DNA and 
mitTF. Additionally, removal of 3 of the 5 C residues found 
immediately downstream from the in vivo L-strand RNA start 
site enabled the bovine mitochondrial transcription activity 



171 
to efficiently initiate at the correct in vivo position in 
vitro . When no Artiodactyl RNA start sites or strong mitTF 
binding domains were available in the template DNA, the 
bovine transcriptional extract was able to initiate RNA 
synthesis from within the nearby T3 promoter region. DNAse 
1 protection analyses show that bovine mitTF activity also 
interacts with sequences within and upstream of this 
bacteriophage promoter region. 

Relatively Small Regions With Relaxed Sequence Preference 
Near Mitochondrial Transcription Start Sites Are Sufficient 
for Initiation In Vitro 

As shown above, deletion of DNA sequences to within 27 
base pairs upstream of the major in vivo L-strand RNA start 
site (within 8 bp of the major in vitro initiation site, on 
templates either without a native initiation site or with 
multiple C residues downstream) had no apparent effect on 
transcription. This result was rather unexpected since the 
deleted region upstream from position -27 contained 
approximately half of the sequences bound and protected by 
the bovine mitTF activity. Specific DNA-binding by mitTF at 
the remaining sequences was relatively weak, giving a barely 
detectable footprint. This demonstrates that although mitTF 
binding is necessary for transcription initiation, strong 
binding to a complete in vitro target sequence is not 
required. 



172 
Upon removal of Artiodactyl RNA start sites and mitTF 
binding domains, in vitro transcription initiation was seen 
outside of the cloned bovine LSP region sequences, at an 
upstream region within the T3 promoter. This demonstrates 
that even without any cognate cis- DNA elements, bovine 
transcriptional extracts are capable of in vitro initiation, 
possibly due to the relatedness recently noted between 
mitRNA polymerases and T3 and T7 RNA polymerases (Masters et 
al. 1987) (but see below) . 

These studies also lead to the proposal that 
mitochondrial RNA polymerase appears to have a hierarchical 
order of preference for initiation sites in vitro. When 3 C 
residues were removed from the 3 ■ end of the LSP region 
(immediately downstream from the major in vivo RNA start 
sit) , in vitro initiation shifted away from the -19 site to 
the native site at +1. When the major in vivo site was 
either removed or inactivated (apparently by the presence of 
multiple downstream C residues) , the primary site of RNA 
synthesis was 19 nucleotides further upstream. This 
secondary initiation region has a limited sequence 
similarity to the native in vivo start site ( CTTAAATA for +1 
vs. CTTTTAAC for -19, where the site of initiation is 
underlined) and was probably selected for that reason. As 
noted in Chapters 3 and 4, the -19 position also serves as a 
minor site of in vivo RNA synthesis. Thus, the -19 site can 
be considered a cryptic initiation site, used sparingly in 



173 
vivo and predominantly in vitro due to inhibitory effects of 
the poly-C domain near position +1. Removal of this -19 
site in templates where the +1 site was inactive resulted in 
initiation within the upstream T3 promoter. This 
demonstrates a third step in the hierarchy of start site 
selection; when neither major nor minor in vivo start sites 
are available, bovine transcriptional extracts will initiate 
near where mitTF has bound. 

A Unified Picture of Vertebrate Mitochondrial Transcription 

On the surface, the results of in vitro transcription 
experiments using the bovine and porcine mitochondrial 
activities appear to be in conflict with those reported for 
other mammalian species. In both the human and mouse, 
deletion of upstream LSP sequences to position -28 relative 
to the RNA start site were associated with significant 
reduction in the efficiency of transcription. We did not 
observe this effect in Artiodactyls. Several lines of 
evidence, however, suggest that previous results and ours, 
here, lead to a unified picture of mitochondrial 
transcription in all vertebrate species. Furthermore, it is 
possible through a compilation of results in various 
vertebrates to deduce the critical common features of 
mitochondrial transcription. 

As shown in Chapter 5, the bovine mitochondrial RNA 
polymerase is dependent upon the presence of mitTF activity 






174 
for specific initiation. The bovine mitTF activity, while 
able to recognize and bind to sequences upstream from 
mitochondrial transcriptional start sites, exhibits a 
distinctly relaxed sequence specificity in its interaction 
with DNA at these regions. In addition, mitTF recognizes a 
variety of sequences at mitochondrial promoters in different 
species. We also note that several areas of DNAse 1 
protection and hypersensitivity were seen outside of the 
normal promoter regions. This relative lack of DNA-binding 
specificity suggests that in the absence of preferred 
homologous recognition sites mitTF will bind less 
specifically to many DNA sequences. Thus, relaxed and 
relatively nonspecific mitTF: DNA interactions probably 
allowed cloned DNA templates with deletions of the bovine 
LSP mitTF binding site (pPAB, and pCS-CD) to remain 
transcriptionally active. The existence of relatively non- 
specific mitTF: DNA interaction is further supported by the 
observation that in footprint assays after incubation with 
the bovine mitTF activity, DNAs without homologous binding 
domains required as much as 4 to 100 times the level of 
DNAse 1 for partial digestion. Additionally, in band shift 
assays (under conditions similar to transcription reactions 
with identical amounts of added bovine mitochondrial protein 
fraction), 0.5 ng of probe DNAs without specific mitTF 
binding sites were fully shifted even in the presence 2 ug 
of unlabeled DNA template. These findings suggest that the 



175 
mitochondrial protein extracts isolated from bovine and 
porcine tissue contain high levels of mitTF activity, 
capable of binding many DNA sequences. Therefore, when 
incubated with DNA templates without strong mitTF 
recognition sites, DNA binding occurs relatively non- 
specif ically. The direct inference is then, that in the 
presence of sufficient levels of mitTF binding, RNA 
polymerase will initiate wherever it finds a suitable nearby 
initiation site(s). 

The concept that the level of mitTF is critical in 
regulating the specificity of transcription initiation can 
be used to develop a unified picture of vertebrate 
mitochondrial promoter structure. This idea can be 
additionally extended to include a role for mitTF in 
determining whether mitDNA in a cell will preferentially 
undergo DNA replication primed by RNA initiated at the LSP 
or preferentially be transcribed. A key common feature of 
the promoter structure in mammalian mitochondria is the 
preferential in vitro binding of the cognate mitTF for its 
LSP rather than its HSP. This has been shown to be true for 
the human, mouse and cow. In human and mouse, strong mitTF 
binding at LSP correlates directly with the apparent 
strength of in vitro transcription. In the human system 
where the mitochondrial RNA polymerase and mitTF have been 
effectively separated and purified, reconstituted specific 
transcriptional activity is significantly more sensitive to 



176 
the deletion of mitTF binding domains at low levels of added 
mitTF activity (Fisher et al. 1987). Conversely, upon the 
addition of much higher levels of mitTF, transcription 
efficiency at mitTF dependent initiation sites is 
effectively reduced. Therefore, at low levels, mitTF will 
associate preferentially with the LSP binding domain, 
thereby promoting transcription from the LSP. Because 
transcription at LSP regions provides the RNA primer for H- 
strand DNA synthesis, the initial event in replicating the 
mitochondrial genome, low mitTF levels will predispose 
mitDNA to be replicated rather than transcribed. At higher 
levels of mitTF, weak binding sites (HSP and potentially 
other sites throughout the mitDNA) will be occupied, thereby 
promoting transcription of the entire genome in both 
polarities. Thus, it is possible that mitTF levels may play 
a critical regularity role in mitochondrial gene expression. 
Since mitTF is a nuclear-encoded protein, regulation of 
mitochondrial replication versus transcription via mitTF 
levels would be controlled by cellular response elements 
with the nucleus. Nuclear regulation of mitochondrial 
function has been well established in simple eukaryotes 
(reviewed in Attardi and Schatz 1989) and the mitTF model 
proposed here creates the analogous regulating linkage in 
higher eukaryotes. This hypothesis makes several testable 
predictions which are discussed and preliminarily assayed in 
Chapter 7. 



177 

Initiation of In Vitro Transcription at the Major In Vivo 
Start Site 

Although numerous different methods for purifying 

transcriptional activity and conditions of in vitro 

transcription were tried, the bovine transcription activity 

would not efficiently initiate in vitro transcription at the 

major +1 in vivo L-strand RNA start site if the template 

DNA contained normal sequences surrounding +1. As discussed 

in Chapter 4 the inability of the bovine RNA polymerase to 

initiate in vitro at the native site is probably not due to 

the absence of a required protein factor (s) from our 

preparation. Although such a situation remains possible, 

the ability of both the bovine and porcine mitochondrial RNA 

polymerase preparations to recognize the correct in vivo LSP 

site on the porcine template strongly argues that all the 

components necessary for accurate transcription are present 

in the extracts. Only when 3 C residues at downstream 

positions +5 to +7 were deleted from the DNA template would 

the in vitro system mimic in vivo initiation. Similar to in 

vitro initiation at the -19 site, initiation from the major 

in vivo site at +1 did not require strong upstream DNA 

binding by the mitTF activity. A major question is then, 

why is this run of C residues apparently inhibitory to 

nearby transcription initiation? The correlation between 

the prominent runs of C residues in the CSB-2+3 domain 

immediately downstream from the L-strand RNA start sites in 

the cow, water buffalo and giraffe, and the 5' end of the D- 



178 
loop DNA in these species, suggests an answer: In vivo , the 
nearby triple stranded D-loop DNA region may influence 
initiation of RNA synthesis at the major site. Since 
truncation of the C run enables initiation to occur 
immediately upstream, we suggest that the primary effect of 
this 3 base deletion was to lower the local T m of this 
domain. If so, in vivo the pre-existing D-loop could 
provide this function for transcription by assisting in the 
local melting of the DNA strands at the site of initiation. 
The presence of a triple stranded D-loop would favor nearby 
transient double stranded DNA melting because it would 
provide a permanent nearby site from which to initiate local 
DNA breathing. At first sight, the location of the porcine 
D-loop DNA 5' end up to several hundred base pairs 
downstream from the L-strand RNA start site seems 
inconsistant with this idea because it would not be expected 
to enhance melting at the LSP. In fact, however, it is a 
good test of the idea. Unlike all other Artiodactyls, 
multiple Cs are not found downstream from the porcine 
transcriptional start site, hence the porcine LSP may not 
need D-loop assisted strand separation. Therefore, the 
porcine seguence arrangement near LSP, by its difference 
from all other Artiodactyls, actually helps confirm the 
hypothesis. Additionally, we note that in vitro 
transcription was more active on supercoiled DNAs. 
Negatively supercoiled DNAs are known to undergo much more 



179 
frequent transient denaturation of complimentary DNA 
strands. Therefore, although sites of initiation were not 
altered in supercoiled templates, the higher transcriptional 
efficiency on supercoiled templates indicates that the 
topological state of the DNA can alter the efficiency of the 
transcriptional process also, possibly by assisting in 
melting DNA strands at L-strand start sequences. 

Transcription Initiation and mitTF Interaction at the T3 
Promoter Region 

The RNA polymerase from yeast mitochondria has 

recently been cloned and sequenced (Masters et al. 1987) . 

This protein exhibits striking amino-acid homology with the 

DNA-dependent RNA polymerases from bacteriophages T3 and T7 

and not with bacterial RNA polymerases. In run-off 

transcription experiments, using templates lacking a 

mitochondrial initiation site, the bovine transcriptional 

activity efficiently initiated RNA synthesis from a site 

within the T3 promoter region, approximately 4 nt downstream 

from the RNA start site used by the T3 polymerase. 

Additionally, the bovine mitTF activity was able to 

recognize and bind to sequences within the T3 promoter 

region. The ability of these bovine mitochondrial 

activities to recognize the T3 promoter suggests that the 

similarity seen at the amino-acid sequence level between 

yeast mitRNA polymerase and the T3 RNA polymerase extends 

beyond catalytic domain similarity to a functional 



180 
similarity at promoter recognition domains as well. In 
other words, it appears that the function of the single T3 
RNA polymerase molecule to both recognize its promoter and 
synthesize RNA are provided by two polypeptides in the 
mitochondria, the mitRNA polymerase and the mitTF. Assuming 
a eubacterial origin for mitochondria (Gray 1989) , the 
evolutionary inference is that after a common origin for T3 
and mitochondrial transcriptional machinery, mitochondria 
have evolved into a two protein system for transcription 
(RNA polymerase and cognate transcription factor) while the 
bacteriophage T3/T7 maintained a single protein system in 
which the catalytic function and seguence recognition 
function remained on the same polypeptide. Of course, 
whether the presumptive common ancestor had a one protein or 
two protein transcriptional apparatus cannot be determined 
from this data. 






CHAPTER 7 
SUMMARY AND PERSPECTIVES 






In an attempt to better define the DNA sequence 
elements necessary for the regulation of mitochondrial RNA 
synthesis, we have studied L-strand transcription in a 
series of related species within the Artiodactyl order (cow, 
water buffalo, giraffe, and pig) . Initially, we mapped the 
L-strand RNA start sites of these species and compared the 
DNA sequences upstream. Immediately encompassing these RNA 
start sites was the 8 bp domain, CTTAAATA, strictly 
conserved in all four species. Upstream from this domain 
several regions of less rigid conservation were also 
observed. In order to test the functional relevance of 
these conserved domains in the regulation of RNA synthesis, 
we developed in vitro transcription systems from the 
mitochondria of the pig and cow (two of the most divergent 
Artiodactyl species) . This is the first report of the 
isolation of transcriptionally active mitochondrial extracts 
purified from normal animal tissue. Although significant 
sequence differences were noted in promoter domains upstream 
from the porcine and bovine L-strand RNA start sites, the 
two transcriptional preparations were functionally 

181 






182 
interchangeable, both being able to accurately transcribe 
either the porcine or bovine template. Extending these 
studies, we were able to identify and partially separate a 
mitochondrial transcription factor activity from the bovine 
RNA polymerase activity. Using DNAse 1 protection assays, 
sites of specific bovine mitTF:DNA interaction were seen 
upstream from both the bovine light and heavy strand RNA 
start sites, as well as at the light and heavy strand 
promoters of the human and mouse, and the light strand 
promoter of the pig. A seguence alignment of regions bound 
by the factor suggests that only rather general DNA sequence 
motifs are required for specific recognition. A synthetic 
Artiodactyl LSP region was designed and made based on the 
evolutionally conserved domains and separated by restriction 
enzyme sites, thus forming movable DNA cassettes. We then 
tested the ability of the bovine mitochondrial transcription 
machinery to interact with a series of mutant LSP regions in 
in vitro transcription and DNAse 1 protection experiments. 
Unexpectedly, the bovine transcriptional activity was able 
to efficiently transcribe DNA templates in which the site of 
mitTF binding had been deleted. When all sites of 
transcription initiation within the nearby LSP region were 
deleted, transcription was detected initiating within the T3 
promoter as well as at other sites. These experiments 
suggest a dynamic relationship between mitTF and RNA 
polymerase, where at low levels of mitTF, transcription is 



183 

limited to regions of high affinity binding, however, at 
increased mitTF levels, factor binding occurs at sites of 
lower affinity, allowing the RNA polymerase to select 
multiple sites of initiation. The results further suggest 
that the balance existing between initiation at D-loop sites 
important for priming DNA replication and initiation at 
other regions in the mitochondrial genome leading to genome 
transcription may be maintained by the level of mitTF 
available within the mitochondrion. 

Does Initiation of Mitochondrial Transcription Occur Outside 
the D-loop Region? 

The consistent theme of these studies has been that 

initiation of mitochondrial transcription is not a rigidly 

promoter seguence controlled process. The relaxed seguence 

recognition of the mitTF and the ability of the 

mitochondrial RNA polymerase activity to initiate RNA 

synthesis at regions of both strong and weak mitTF DNA 

binding suggests that RNA synthesis initiation may not be 

limited to the D-loop region but may also emanate from a 

number of regions throughout the mitochondrial genome. It 

has been previously proposed that the compact nature of the 

vertebrate mitochondrial genome, such that there is little 

or no non-coding sequence between genes, severely reduces 

the likelihood of transcription initiation outside the D- 

loop region (Clayton 1984) . However given the possibility 

of intragenic transcription initiation, the extremely 



184 
compact organization of the genome could actually be an 
advantage. Mitochondrial tRNA genes are situated between 
almost every coding region. Since an active system of RNAse 
P-like activities is known in mitochondria, and the 
secondary structure of these RNAs is thought to serve as a 
signal for such RNA processing, transcripts initiating 
within a protein coding region would soon encounter multiple 
processing signals thus providing a mechanism for efficient 
removal of the incomplete leader RNA and release of mature 
gene length transcripts. Transcription initiation within a 
tRNA gene would merely result in the addition of less than a 
hundred bases of nontranslated leader to a downstream mRNA. 

As mentioned in Chapter 5, the ability of the mitTF to 
bind specifically to a broad spectrum of DNA seguences 
suggests that binding of this protein may occur at several 
sites in the mitochondrial genome outside the D-loop region. 
Indeed, a comparison of all regions bound by the bovine 
mitTF with seguences throughout the bovine genome identifies 
several areas of seguence reasonable similarity (Figure 7- 
1) . Interestingly, several of the regions identified are 
contained within tRNA genes. 

In preliminary attempts to determine if these regions 
of putative mitTF interaction are capable of directing RNA 
synthesis in vivo and in vitro , we synthesized a DNA 
oligonucleotide complementary to seguences near a cluster of 
tRNA genes for use in primer extension assays. A cloned DNA 






Figure 7-1. Sequence alignment of regions of the bovine 
mitochondrial genome with similarity to the bovine LSP region . 
Sequences throughout the bovine genome, with limited 
similarity to the mitTF binding site at the bovine LSP region 
were aligned. The spacing used to divide sequence motifs in 
Figure 5-10, Panel D was also used here. Names preceding 
sequences are the corresponding coding regions where the 
regions reside. 












186 



BOVINE 


LSP 


CTTTTAACAG 


ACTTTTCCC 


TAGATAC 


ND 5 




GCATTAACCA 


ACCTTACC 


TAGCTTT 


12S 


rRNA 


ATTTTAATC 


ATGGCTTTT 


TACAGCT 


HISTIDINE 


tRNA 


ACTTTAAAAA 


AACAT 


TAG TCA 


SERINE 


tRNA 


TTTTTCGAAC 


TTTTAAAGGA 


TAG TAG 


ISOLEUCINE 


tRNA 


CTTTGA TAC 


ACTAAATAAA 


TAGAGG 


16S 


rRNA 


CTTTTAAT 


CTTTCCT 


TAGATGC 


GLYCINE 


tRNA 


CTTTTAGTAT 


TAAC 


TAG TAC 


ARGININE 


tRNA 


AGTTTAAAAT 


AAAATAAA 


T GATTC 


PROLINE 


tRNA 


TATTTAAACT 


ATTCCC 


TGAACAC 



187 
containing these tRNA genes was then incubated in an in 
vitro transcription reaction and RNA products mapped by 
primer extension. Primer extension of in vivo RNA was 
performed in parallel in an attempt to correlate in vivo 
mitochondrial RNA 5 ■ ends with sites of in vitro 
transcription initiation. As shown in Figure 7-2, two 
prominent sites of in vitro transcription initiation are 
seen in lane 1 corresponding to positions within the 
histidine tRNA gene. Primer extension of the in vivo mitRNA 
reveals the presence of two labeled species having 5 ' ends 
very near the in vitro product (lane 2) . However, other 
labeled species of equal intensity were also seen in several 
other locations. At this point, it is impossible to 
determine if these various in vivo RNA 5 ' ends represent 
primary RNAs or are simply the result of RNA processing 
and/or degradation. However, the apparent efficiency with 
which this region is transcribed in vitro (apparently 
equivalent to in vitro transcription at LSP) adds support to 
our hypothesis and indicates that a more detailed analysis 
is justified. In order to fully define mitochondrial 
transcription from non-D-loop regions, these experiments 
should include a combination of in vitro capping of the in 
vivo RNA, to correlate with in vitro transcripts. 
Additionally, in vitro initiation predicts specific 
mitTF:DNA interaction and this must be experimentally 
verified. A second prediction of the mitTF model for 



Figure 7-2. In vivo and in vitro transcription in a non-D- 
loop region of the bovine mitDNA . A cloned DNA containing 
bases 11692-12402 of the. bovine genome (which included a 
cluster of 3 mitochondrial tRNA genes from nucleotides 11907- 
12108) was incubated with the bovine mitochondrial 
transcription extract. The RNA products were then annealed 
to a radiolabeled DNA oligonucleotide, corresponding to 
nucleotides 11846-11867 on the bovine mitDNA, and primer 
extended with reverse transcriptase (lane 1) . In parallel, 
an aliquot of about 10 ug of mitRNA (previously DNAse 1 
treated as described in Materials and Methods) was also primer 
extended using the same labeled oligonucleotide (lane 2) . 
Lanes A, G, C, and T indicate dideoxy sequencing reactions of 
the corresponding DNA template. The positions of the 
histidine, serine, and leucine tRNA genes are as indicated. 
The nucleotide sequence at the bottom indicates the major 5' 
ends of RNAs synthesized in vitro (top) and in vivo (bottom) . 



189 



A G C . T 1 2 




IN VITRO <\At W 

CAlTAGAriUreMTCTMCMTAGAMCTCATTACTrCTr 

IN VIVO ft f k t 



190 
balancing mitDNA replication and transcription is that in 
actively dividing cells relatively low levels of mitTF 
should be present in mitochondria and that LSP should be 
preferentially bound. In contrast, non-dividing, but 
metabolically active cells should have high levels of mitTF 
and consequently less preference for binding to LSP. This 
prediction may be amenable to testing in bovine tissue 
culture cells by comparing log phase cells versus 
stationary, confluent cells. In vivo footprinting of LSP in 
each cell phase would seem to be the most direct 
experimental test of this prediction. 












LIST OF REFERENCES 



Aloni, Y. and G. Attardi. 1971a. Expression of the 
mitochondrial genome in HeLa cells II. Evidence for complete 
transcription of mitochondrial DNA. J. Mol. Biol. 55:251- 
267. 

Aloni, Y. and G. Attardi. 1971b. Symmetrical in vivo 
transcription of mitochondrial DNA in HeLa cells. Proc. Natl. 
Acad. Sci. U.S.A. 68:1757-1761. 

Anderson, S., Banker, A. T. , Barrell, B. G. , de Bruijin, M. 
H. L., Coulson, A. R. , Ovouin, J., Eperon, I. C. , Nierlich, 
D. P., Roe, B. A., Sanger, F. , Shreier, P. H. , Smith, A. J. 
H. , Staden, R. and I. C. Young. 1981. Sequence and 
organization of the human mitochondrial genome. Nature 
290:457-465. 

Anderson, S., de Bruijn, M. H. L. , Coulson, A. R. , Eperon, E. 
C, Sanger, R. and I. G. Young. 1982. Complete sequence of 
bovine mitochondrial DNA: conserved features of the mammalian 
mitochondrial genome. J. Mol. Biol. 156:683-717. 

Bennett J. L. , and D. A. Clayton. 1990. Efficient site- 
specific cleavage by RNase MRP requires interaction with two 
evolutionarily conserved mitochondrial RNA sequences. Mol. 
Cell. Biol. 6:409-417. 

Bibb, M. J., Van Etten, R. A., Wright, C. T. , Walberg, M. W. 
and D. A. Clayton. 1981. Sequence and gene organization of 
the mouse mitochondrial DNA. Cell 26:167-180. 

Biswas, T. K. , Edwards, J. C. , Rabinowitz, and G. S. Getz . 
1985. Character izatin of a yeast mitochondrial promoter by 
deletion mutagenesis. Proc. NAtl. Acad. Sci. U.S.A. 
82:1954-1958. 

Bogenhagen, D. F., Applegate, E. F. and B. K. Yoza. 1984. 
Identification of individual promoters for transcription of 
the heavy strand of human mtDNA: in vitro transcription and 
deletion mutagenesis. Cell 36:1105-1113. 

Bogenhagen, D. F. and N. F. Insdorf. 1988. Purification of 
Xenopus laevis mitochondrial RNA polymerase and identification 
of a dissociable factor required for specific transcription. 
Mol. Cell. Biol. 8:2910-2916. 



191 



192 

Bogenhagen, D. F. and M. F. Romanelli. 1988. Template 
sequences required for transcription of Xenopus laevis 
mitochondrial DNA from two bidirectional promoters. Mol. 
Cell. Biol. 8:2917-2924. 

Bogenhagen, D. F. and B. K. Yoza. 1986. Accurate in vitro 
transcription of Xenopus laevis mitochondrial DNA from two 
bidirectional promoters. Mol. Cell. Biol. 6:2543-2550. 

Bogenhagen, D. F., Yoza, B. K., and S. S. Cairns. 1986. 
Identification of initiation sites for transcription of 
Xenopus laevis mitochondrial DNA. J. Biol. Chem. 261:8488- 
8494. 

Cantatore, R. and G. Attardi. 1980. Mapping of nascent light 
and heavy strand transcripts on the physical map of HeLa cell 
mitochondrial DNA. Nucleic Acids Res. 8:2605-2625. 

Chang, D. D. and D. A. Clayton. 1984. Precise identification 
of individual promoters for transcription of each strand of 
human mitochondrial DNA. Cell 36:635-643. 

Chang, D. D. and D. A. Clayton. 1985. Priming of human 
mitochondrial DNA replication occurs at the light strand 
promoter. Proc. Natl. Acad. Sci. U.S.A. 82:351-355. 

Chang, D. D. and D. A. Clayton. 1986a. Identification of 
primary transcriptional start sites of mouse mtDNA: accurate 
in vitro initiation of both heavy and light strand 
transcripts. Mol. Cell. Biol. 6:1446-1453. 

Chang D. D. and D. A. Clayton. 1986b. Precise assignment of 
the L-strand promoter of mouse mitochondrial DNA: a functional 
promoter consists of multiple upstream domains. Mol. Cell. 
Biol. 6:3252-3261. 

Chang. D. D. and D. A. Clayton. 1986c. Precise assignment 
of the heavy strand promoter of mouse mitochondrial DNA: 
cognate start sites are not required for transcriptional 
initiation. Mol. Cell. Biol. 6:3262-3267. 

Chang, D. D. , and D. A. Clayton. 1987. A novel 
endoribonuclease cleaves at a priming site of mouse 
mitochondrial DNA replication. E.M.B.O. 6:409-417. 

Chang, D. D. Hauswirth, W. W. , and D. A. Clayton. 1985a. 
Replication priming and transcription initiate from precisely 
the same site in mouse mitochondrial DNA. E.M.B.O. 4:1559- 
1567. 

Chang, p. D., Hixson, J. E. and D. A. Clayton. 1986. Minor 
transcriptional initiation events indicate that both human 



193 

mitochondrial promoters function bidirectionally. Mol. Cell. 
Biol. 6:294-301. 

Christianson, T. W. and D. A. Clayton. 1986. In vitro 
transcription of human mitochondrial DNA: accurate termination 
requires a region of DNA sequence that can fucntion 
bidirectionally. Proc, Natl. Acad. Sci. U.S.A. 83:6277-6281. 

Christianson, T., Edwards, J. C, Mueller, D. M. , and 
Rabinowitz, M. 1983. Identification of a single 
transcriptional initiation site for the glutamic tRNA and COB 
genes in yeast mitochondria. Proc. Natl. Acad. Sci. U.S.A. 
80:5564-5568. 

Christianson, T. and M. Rabinowitz. 1983. Identification of 
multiple transcriptional initiation sites on the yeast 
mitochondrial genome by in vitro capping with guanylyl 
transferase. J. Biol. Chem. 258:14025-14033. 

Clayton, D. A. 1984. Transcription of the mammalian 
mitochondrial genome. Ann. Rev. Biochem. 53:573-594. 

Doda, J. N., Wright, D. T. and D. A. Clayton. 1981. 
Elongation of displacement-loop strands in human and mouse 
mitochondrial DNA is arrested near specific template 
sequences. Proc. Natl. Acad. Sci. U.S.A. 78:6116-6120. 

Fisher, R. P. and D. A. Clayton. 1985. A transcription 

factor required for promoter recognition by human 

mitochondrial RNA polymerase. J. Biol. Chem. 260:11330- 
11338. 

Fisher, R. P., and D. A. Clayton. 1988. Purification and 
Characterization of human mitochondrial transcription Factor 
1. 8:3496-3509. 

Fisher, R. P., Parisi, M. A. and D. A. Clayton. 1989. 
Flexible recognition of rapidly evolving promoter sequences 
by mitochondrial transcription factor 1. Genes and Dev. 
3:2202-2217. 

Fisher R. P., Topper, J. N., and D. A. Clayton. 1987. 
Promoter selection in human mitochondria involves binding of 
a transcription factor to orientation-independent upstream 
regulatory regions. Cell 50:247-258. 

Gaines, G. , and G. Attardi. 1984. Highly efficient RNA- 
synthesizing system that uses isolated human mitochondria: new 
initiation events and in vivo -like processing. Mol. Cell 
Biol. 4:1605-1617. 



194 

Geliebter, J. 1987. Dideoxynucleotide sequencing of RNA and 
uncloned cDNA. Focus 9:5-7. 

Gillum, A. and D. A. Clayton. 1979. Mechanism of 
mitochondrial DNA replication in mouse L-cells: RNA priming 
during the initiation of heavy-strand synthesis. J. Mol. 
Biol. 135:353-368. 

Gray, M. W. 1989. Origin and evolution of mitochondrial DNA. 
Ann. Rev. Cell. Biol. 5:25-50. 

Hixson, J. E. and D. A. Clayton. 1985. Initiation of 
transcription from each of two human mitochondrial promoters 
requires unique nucleotides at the transcriptional start 
sites. Proc. Natl. Acad. Sci. U.S.A. 82:2660-2664. 

Hess, J. F., Parisi. M. A., Bennett, J. L. and D. A. Clayton. 
1991. Impairment of mitochondrial transcription termination 
by a point mutation associated with the MELAS subgroup of 
mitochondrial encephalomyopathies. Nature. 351:236-239. 

Jacob, H. T., Elliot, D. J., Moth, V. B. and A. Farguharson. 
1984. Nucleotide sequence and organization of sea urchin 
mitochondrial DNA. 202:185-217. 

King, T. C, and R. L. Low. 1987. Mapping of control 
elements in the displacement loop region of mitochondrial DNA. 
J. Biol. Chem. 262:6204-6213. 

Kruse, B., Narasimhan, N., and G. Attardi. 1989. Termination 
of transcription in human mitochondria: identif iaction and 
purification of a DNA binding protein factor that promotes 
termination. Cell 58:391-397. 

Levens, D., Lustig, A., and M. Rabinowitz . 1981. 
Purification of mitochondrial RNA polymerase from 
Sacchar omvces cerevisiae . J. Biol. Chem. 256:1474-1481. 

Low, R. L., Cummings, O. W. , and T. C. King. 1987. The 
bovine mitochondrial endonuclease prefers a conserved sequence 
in the displacement loop region of mitochondrial DNA. J. 
Biol. Chem. 262:16164-16170. 

Masters, B. S., Stohl, L. L. , and D. A. Clayton. 1987. Yeast 
mitochondrial RNA polyermase is homologous to those encoded 
by bacteriophages T3 and T7 . Cell 51:89-99. 

Maxam, A. M. , and W. Gilbert. 1980. Sequencing end-labeled 
DNA with base-soecific chemical cleavage. Methods Enzymol. 
65:499-560. 






195 

Montoya, J., Christianson, T. , Levens, D., Rabinowitz, M. and 
G. Attardi. 1982. Identification of initiation sites for 
heavy strand and light strand transcription in human 
mitochondrial DNA. Proc. Natl. Acad. Sci. U.S.A. 79:7195- 
7199. 

Montoya, J., Gaines, G. L. and G. Attardi. 1983. The pattern 
of transcription of the human mitochondrial rRNA genes reveals 
two overlapping transcription units. Cell 34:151-159. 

Murphy, W. I., Attardi, B., Tu, C. and G. Attardi. 1975. 
Evidence for complete symmetrical transcription in vivo of 
mitochondrial DNA in HeLa cells. J. Mol. Biol. 99:809-814. 

Parisi, M. A. and D. A. Clayton. 1991. Similarity of human 
mitochondrial transcription factor 1 to high mobility group 
proteins. Science. 252:965-969. 

Roe, B. A., Ma, D.-P, Wilson, R. K. and J. F. -H. Wong. 1985. 
The complete nucleotide sequence of the Xenopus laevis 
mitochondrial genome. J. Biol. Chem. 260:9759-9774. 

Schinkel, A. H. , Groot Koerkamp, M. J. A., Touw, E. P. W. , and 
H. F. Tabak. 1987. Specificity factor of yeast mitochondrial 
RNA polymerase. J. Biol. Chem. 262:12785-12791. 

Schinkel, A. H. , Groot Koerkamp, M. J. A., and H. F. Tabak. 
1988. Mitochondrial RNA polymerase of Saccharomyces 
cerevisiae: compososition and mechanism of promoter 
recognition. E.M.B.O. 7:3255-3262. 

Shuey, D. J. and G. Attardi. 1985. Characterization of an 
RNA polymerase activity from HeLa cell mitochondria which 
initiates transcription at the heavy strand rRNA promoter and 
the light strand promoter in human mitochondrial DNA. J. 
Biol. Chem. 260:1952-1958. 

Walberg, M. W. and D. A. Clayton. 1983. In vitro 
transcription of human mitochondrial DNA. Identification of 
specific light-strand transcripts from the displacement-loop 
region. J. Biol. Chem. 258:1268-1275. 

Williams, T., and R. Tijan. 1991. Characterization of a 
dimerization motif in AP-2 and its function in heterologous 
DNA-binding proteins. Science. 251:1067-1071. 

Wong, T. W. and D. A. Clayton. 1985a. In vitro replication 
of human mtDNA: accurate initiation at the origin of L-strand 
synthesis. Cell 42: 951-958. 



196 

Wong, T. W. and D. A. Clayton. 1985b. Isolation and 
characterization of a DNA primase from human mitochondria. 
J. Biol. Chem. 21:11530-11535. 

Young, P. G. and G. Attardi. 1975. Characterization of 
double-stranded RNA from HeLa cell mitochondria. Biochem. 
Biophys. Res. Comm. 65:1201-1207. 

Yoza, B. K. and D. F. Bogenhagen. 1984. Identification and 
in vitro capping of a primary transcript of human 
mitochondrial DNA. J. Biol. Chem. 259:3909-3915. 












BIOGRAPHICAL SKETCH 






Steven C. Ghivizzani was born in Hampton, Virginia on 
September 13, 1958. At the age of five, he moved to Orlando, 
Florida with his mother Carolyn, sister Darling, and brother 
Scot. He graduated from Colonial High School in the spring 
of 1976 and in the fall entered the University of Central 
Florida. During undergraduate studies he was employed at Walt 
Disney World where he worked for five years. While working 
in the Tickets Department at Disney World he met his future 
wife, Nancy K. Murray. In 1981 he graduated with a B. A. in 
Psychology. For the next three years he took a hiatus from 
school, during which he had a part-time job with United Parcel 
Service allowing him time to pursue interests in music and 
water skiing. In 1984 he entered the graduate program of the 
Department of Immunology and Medical Microbiology at the 
University of Florida where he remained until the present 
time. In December 1988, he was married to Nancy Murray and 
shortly thereafter was blessed with two beautiful daughters, 
Meredith and Alisyn. 



I9> 



I certify that I have read this study and that in my 
opinion it conforms to acceptable standards of scholarly 
presentation and is fully adequate, in scope amd quality, as 
a dissertation for the degree of Doctor of Philosophy. 





William W. Hauswirth, Chairman 
Professor of Immunology and 
Medical Microbiology 



I certify that I have read this study and that in my 
opinion it conforms to acceptable standards of scholarly 
presentation and is fully adequate, in scope and quality, as 
a dissertation for the degree of Doctor of Philosophy. 

Donna H. Duckworth 
Professor of Immunology and 
Medical Microbiology 

I certify that I have read this study and that in my 
opinion it conforms to acceptable standards of scholarly 
presentation and is fully adequate, in scope and quality, as 
a dissertation for the degree of Doctor of Philosophy. 



Henry V. Baker 
Assistant Professor of 

Immunology and Medical 

Microbiology 



I certify that I have read this study and that in my 
opinion it conforms to acceptable standards of scholarly 
presentation and is fully adequate, in scope and quality, as 
a dissertation for the degree of Doctor of Philosophy. 



Edward K. Wakeland 
Associate Professor of Pathology 
and Laboratory Medicine 



This dissertation was submitted to the Graduate Faculty 
of the College of Medicine and to the Graduate School and was 
accepted as partial fulfillment of the requirements for the 
degree of Doctor of Philosophy. 



December, 1991 ^j ij ^yy ?/^ 





Dean, College of Medicine 



n 



Dean, Graduate School 



UNIVERSITY OF FLORIDA 

HP